thanatchon36/ai-zeroday-banking

Zero-Days at Machine Speed: dataset

Data behind the paper Zero-Days at Machine Speed: AI-Driven Vulnerability Discovery, Systemic Cyber Risk in Banking, and Implications for Central Banks (Thanatchon Chongmankhong, 2026).

• 📖 Paper online: https://thanatchon36.github.io/ai-zeroday-banking/paper/
• 💻 Code: https://github.com/thanatchon36/ai-zeroday-banking
• 🧮 Simulation model card: https://huggingface.co/thanatchon36/ai-zeroday-banking

Contents

Path	Rows	What it is
processed/cve_ai_flags.csv.gz	405,375	Every published CVE (CVE List V5, snapshot 2026-11-12) with vendor, product, CWE, max CVSS, SSVC exploitation, and AI-credit flags (ai_tier 1 = AI system named, 2 = AI-native firm)
processed/ai_credited_cves.csv	1,598	The AI-credited subset with credit text and description
processed/kev_panel.csv	1,779	CISA KEV joined to CVE dates: lag, category, banking-relevance weight, ransomware use, AI flag
processed/sim_draws.csv.gz	—	Monte Carlo draws per scenario from the Thai-like banking system simulation
curated/	small	Hand-built inputs: AI-credit lexicon and exclusions, KEV category rules, banking weights, 33 financial-sector incidents, hardware trust-anchor cases, AI milestones, Thai group scores, zero-day benchmarks
tables/	small	Every table in the paper as CSV
key_numbers.json	—	Every number quoted in the paper

Sources and licence

Derived from public data: CVE List V5 (CVE Program terms of use) and CISA KEV (U.S. government work). Curated files and derived tables: CC BY 4.0.

Caveats

• AI counts are lower bounds: only part of CVEs carry credit fields (43% in 2026).
• KEV records exploitation known to CISA, not all exploitation.
• The curated files (financial_incidents.csv, hardware_trust_anchors.csv, ai_milestones.csv, zero_day_benchmarks.csv) were compiled without live web access; verify against primary sources before reuse.
• The Thai analysis is group-level structured judgement; no named institution is assessed.
• No exploit code or proof-of-concept is included.

Citation

@article{chongmankhong2026zerodays,
  title  = {Zero-Days at Machine Speed: AI-Driven Vulnerability Discovery,
            Systemic Cyber Risk in Banking, and Implications for Central Banks},
  author = {Chongmankhong, Thanatchon},
  year   = {2026},
  url    = {https://github.com/thanatchon36/ai-zeroday-banking}
}