One robots.txt line to opt out of AI training, from 2027
The token is called __usage_ai-training, comes from an IETF draft and is not finished. Meanwhile Google-Extended still works and does not affect Google Search rankings.
Jump to a section
Owners of news sites, blogs and online shops want people to find them through Google, but many do not want articles they wrote themselves used to train AI. The problem is that each AI company uses its own crawler name, so blocking them all means writing a block for each one.
That is the problem Google says it will help solve. On 30 October 2026, John Mueller, a Search Advocate at Google, wrote on the Google Search Central blog that Google plans to support a shared signal in robots.txt, the file a website uses to tell crawlers which parts they may visit. The signal comes from a draft by the AIPREF working group at the IETF, the body that drafts internet standards.
Picture a no-entry sign at your front door. Until now you needed a separate sign for each delivery company. Soon there will be one sign that every participating company reads the same way.
The token is __usage_ai-training, and it starts in 2027
Google's crawler documentation says Google intends to support the signal in early 2027, written as user-agent: __usage_ai-training. Mueller's post says Google aims to support it along with other AI providers in 2027 and will give updates on its progress.
Search Engine Roundtable reported that Google said it is following Bing, OpenAI and Anthropic, which have taken similar steps.
Using the token counts as blocking Google-Extended
Google said that if a site uses the token to opt out of training and does not specify Google-Extended, Google will treat it as if the site had disallowed Google-Extended. Each company's existing controls keep working alongside it.
Google-Extended is the token Google uses to separate AI training from search. Google's documentation says it controls whether content Google crawls may be used to train future Gemini models and for grounding answers, and that it does not affect a site's inclusion in Google Search and is not a ranking signal.
The trade-off to know is that sites that allow Google-Extended remain eligible to appear in Gemini app responses, according to Google's post.
The standard is not finished, and there are disputes
The draft is still being worked on at the IETF. Search Engine Roundtable quoted Nate Hake saying significant disputes remain, and that he is unhappy the current draft exempts actions users start themselves. Search Engine Roundtable also noted that Google has a separate Search Console control for AI in Search, apart from this token.
What site owners can do now
- If you do not want Google to use your content to train Gemini, you can use Google-Extended today by adding two lines to robots.txt: User-agent: Google-Extended and Disallow: /
- Keep your existing lines that block other AI companies' crawlers, because the shared token has not started
- Weigh it first, because sites that allow Google-Extended can still be cited in Gemini app answers
- Watch for Google's update in early 2027, then add the __usage_ai-training token once support starts
Open your site's robots.txt today and check who it currently allows or blocks.
FAQ
Will blocking Google-Extended remove my site from Google Search?
No. Google's documentation says Google-Extended does not affect a site's inclusion in Google Search and is not a ranking signal.
Can I use the __usage_ai-training token now?
Google intends to support it in early 2027, and the draft standard is not finished. Until then, use each company's own token.
Which AI companies will read the token?
Google's post refers to other participating AI providers without naming them. Search Engine Roundtable reported that Google said Bing, OpenAI and Anthropic have taken similar steps.
Will the token stop AI summaries of my site in Google Search?
Google's post does not say the token affects Search. Search Engine Roundtable noted that AI in Search has a separate control in Search Console.
