What is an AI training opt-out?
An AI training opt-out is a machine-readable statement that a site's pages should not be used to train AI models, usually set in robots.txt.
Also called __usage_ai-training.
Full guide: The robots.txt AI Training Opt-Out (__usage_ai-training): Syntax, Dates, and Who Honors It
Until October 2026 the AI companies themselves offered only their own controls: GPTBot for OpenAI, ClaudeBot for Anthropic, Google-Extended for Google, Applebot-Extended for Apple. On Oct 30, 2026 an IETF Internet-Draft defined a shared robots.txt policy token, __usage_ai-training. A group naming it with "Disallow: /" states a preference against AI training to every company that reads it. Microsoft says it will honor it from Feb 1, 2027, Google plans support in 2027, OpenAI "in the coming months," and Anthropic says it has started implementing support.
An opt-out is a preference, not a block: it does not stop crawling, and the AIPREF working group lists technical enforcement as out of scope. It is not retroactive: it applies to copies retrieved while it is set, once a company honors it. Bing will also honor a page-level "ai-training:disallow" tag from Feb 1, 2027, and Cloudflare's Content-Signal line offers a separate "ai-train=no" signal.
Why does it matter to your business?
An AI training opt-out lets a business decline AI training without leaving search, as long as search crawlers stay allowed.
Sources, verified October 31, 2026
Want to know how your site measures up?
The free audit checks whether Google and AI answer engines can find, understand, and quote your pages.