The proposed standard for telling AI systems what a site offers. Google says its Search doesn't read it; adoption, tooling, and the other engines are the live questions.
Two items, and between them they close half of the open question. Ahrefs shipped a free llms.txt generator in late July 2026, built on its analysis of adoption across 137,000 domains — tooling that drops the cost of publishing the file to near zero. Then Google's AI-features documentation answered its own half plainly: you don't need to create machine readable files, AI text files, or markup to appear in those features. What the record does not contain is the other half: no statement from OpenAI, Anthropic, Perplexity or Microsoft about whether their systems read the file, and no measurement anywhere tying its publication to citations. Adjacent to it, the wire has Time serving agent-readable markdown pages and selling ads inside them.
5 items on the wire, newest first.
Elsewhere: official · Wikidata
Common Crawl analyzed 584,107 llms.txt files gathered in its July 2026 crawl archive and found 68% were produced by templates or SEO plugins — Wix alone…
An Ahrefs analysis spanning roughly 15 million data points found 97% of sites' llms.txt files were never read by AI crawlers and that adding schema markup…
Jeremy Howard, the Answer.AI developer who created llms.txt, published a version 2 update to the spec on August 10 — its first revision since the format…
Google's "AI features and your website" documentation states plainly: "You don't need to create new machine readable files, AI text files, or markup to…
Ahrefs released a free web tool that builds an llms.txt file for any domain, drawing on the company's prior analysis of llms.txt adoption across 137,000…