README.md

metadata

license: cc-by-4.0
pretty_name: State of AI crawler access, October 2026
language:
  - en
tags:
  - robots-txt
  - llms-txt
  - web-crawling
  - seo
  - tabular
  - csv
size_categories:
  - 1K<n<10K
configs:
  - config_name: domains
    data_files: domains.csv
    default: true
  - config_name: figures
    data_files: figures.csv

State of AI crawler access, October 2026

robots.txt rules for 11 AI crawlers and the catch-all group, and llms.txt presence and structure, for a fixed panel of 9,973 top domains (Tranco list N26XW, first 10,000 places, less 2 invalid entries and 25 domains of companies that build AI models), read on 15 October 2026 by QueryRevBot.

What is in it

• domains.csv: one row per domain of the panel, with what we read (robots.txt status and cause of failure, llms.txt found or not, its size, summary, link sections and links checked) and how each of the 11 crawlers is treated (allowed, partial, blocked).
• figures.csv and figures.json: every figure with its count and three denominators (domains in the panel, domains whose robots.txt we could read, domains whose robots.txt names an AI crawler), overall, by rank band and by country-code ending.

Read this first

6,294 of the 9,973 domains (63.1%) gave a robots.txt we could read; the other 3,679 are marked as not read, not as having no file. llms.txt is counted only when /llms.txt answers 200 with text that opens with a Markdown title. Scope: the 11 crawlers measured, not every AI crawler. robots.txt is a request, not enforcement, and the data comes from one vantage point.

The panel is fixed, so later editions are compared with this one. Full method and limits: https://queryrev.com/research/ai-crawler-access?utm_source=huggingface&utm_medium=social&utm_campaign=nov26-hf-01#method

License and citation

License: CC BY 4.0.

Cite: QueryRev Research. “State of AI crawler access, October 2026.” QueryRev AI, 15 October 2026. https://queryrev.com/research/ai-crawler-access