Yes. AI training and AI answers are controlled by separate settings, so you can opt out of training and stay eligible to show up when ChatGPT, Google's AI Overviews or Bing and Copilot answers suggest a business. That matters because on October 30, 2026, Google announced it plans to support a shared robots.txt line that opts a site out of AI training, starting in 2027. This guide explains what that line does, what it leaves alone, and how to stay visible to AI either way.
What's the difference between AI training and AI answers?
Training is when an AI company uses your pages to build its next model. AI answers are when a tool like ChatGPT search or Google's AI Mode reads your site at question time and cites or links to you. The second one is how AI tools find and recommend a business, and it is controlled by different crawlers.
For most local service businesses, such as plumbers, electricians and pest control companies, the website is marketing. There is usually nothing on it you'd want to keep from AI, and the goal is for every AI tool to know who you are, what you do and where you work. Helping businesses get found in AI answers is the focus of our AI search visibility service. If you'd still rather not have your pages used for training, the new opt-out lets you say so without stepping out of AI answers.
What did Google announce?
Google says it plans to honor a single "policy token" in robots.txt that sets your AI training preference for every provider that supports it, instead of writing a separate rule for each company. The token comes from a proposal (an individual Internet-Draft) submitted to the IETF, the group that writes internet standards, and it looks like this:
User-Agent: __usage_ai-training
Disallow: /Those two lines say "don't use any page on this site to train AI models." Google's crawler documentation says it intends to support the signal in early 2027. Microsoft announced the same day that Bing will start honoring training and grounding preferences on February 1, 2027, and the draft's author, posting on the IETF's mailing list, said Anthropic has started building support, rolling out over the coming months. OpenAI's crawler documentation doesn't mention the token yet (as of November 2026).
Will opting out of AI training stop AI from recommending you?
Not in ChatGPT search, Google's AI Overviews and AI Mode, or Bing and Copilot answers. Your eligibility for each is controlled by a setting the training opt-out doesn't touch:
| Provider | Training control | AI answers control |
|---|---|---|
| Google-Extended, or the shared token from early 2027 | AI Overviews and AI Mode: Googlebot and snippet rules. Gemini Apps: Google-Extended (tied to training) | |
| Bing | The shared token, or a page-level ai-training:disallow meta tag or header (page level overrides robots.txt), from Feb 1, 2027 | Page-level ai-grounding meta tag or header only; no site-wide robots.txt setting |
| OpenAI | GPTBot in robots.txt | OAI-SearchBot in robots.txt; block it and you won't be shown in ChatGPT search answers (navigational links aside) |
At Google, the page on AI features and your website says AI Overviews and AI Mode follow the normal Search controls: robots.txt rules for Googlebot, plus nosnippet and noindex. OpenAI says each of its settings is independent, so a site can block GPTBot and stay eligible for ChatGPT search through OAI-SearchBot. Bing says its new preferences don't change eligibility for regular Bing results, and it warns that blocking grounding means "reduced brand visibility in AI experiences." One thing to check now: Bing currently reads NoArchive and NoCache tags as training and grounding preferences, until it detects the new settings or March 1, 2027. If your site uses them, you may already be out of Bing's grounding.
Your Google rankings don't change either. Google's crawler documentation says Google-Extended "does not impact a site's inclusion in Google Search nor is it used as a ranking signal."
How do you stay eligible for Gemini answers if you opt out?
Name Google-Extended in your robots.txt and allow it.
Google says it will treat the token as a Google-Extended block only when Google-Extended isn't named. Google-Extended covers both training future Gemini models and grounding, which is when Gemini Apps pulls in content from Google's index to answer a question, and Google says sites that specifically allow it (name it in robots.txt) stay eligible to appear in Gemini Apps responses and receive clicks.
So at Google the two go together. Naming and allowing Google-Extended should keep you eligible for Gemini Apps answers (Google hasn't described this exact combination, but it follows from both rules), and it also lets Google use your pages for Gemini training. The example file below does exactly that while opting out of training everywhere else.
What should you do now?
First, make sure nothing in your robots.txt blocks the crawlers that let AI recommend you. Then, if you care about training, decide before early 2027:
- Open yourdomain.com/robots.txt in a browser. Look for any
Disallow: /underUser-Agent: *, Googlebot, Bingbot or OAI-SearchBot. Those block search and AI answers, not just training. Our guide on allowing AI crawlers on your website shows the rules line by line. - Decide whether you want to opt out of training. For most local service businesses, the answer is to stay fully open: leave robots.txt as it is. With no
__usage_ai-traininggroup in the file, no preference is expressed, so each company's existing crawler rules decide. - If you do want out, add the training groups and keep the answer crawlers open. A
User-Agent: *group does not count for the token; the draft says it must be named on its own. You can exempt a folder with anAllowline, such asAllow: /press-kit/. - Keep company-specific rules too. The token only counts once each company supports it, so a GPTBot rule is still how you opt out of OpenAI training today.
Example: opt out of training, stay in AI answers
Only use this if you've decided to opt out. This file opts out of AI training at every company that honors the shared token except Google, and at OpenAI today. Allowing Google-Extended keeps you eligible for Gemini Apps answers but also allows Gemini training. The file leaves Googlebot, Bingbot and OAI-SearchBot free to visit:
# Shared training opt-out (2027)
User-Agent: __usage_ai-training
Disallow: /
# OpenAI training opt-out
User-Agent: GPTBot
Disallow: /
# Stay in Gemini Apps answers
User-Agent: Google-Extended
Allow: /
# Everyone else, incl. search
User-Agent: *
Allow: /Remove the Google-Extended group only if keeping your pages out of Gemini training matters more to you than appearing in Gemini Apps. Keep any Sitemap: line your file already has.
When does the AI training opt-out take effect?
Google says early 2027 in its crawler documentation, and Bing starts February 1, 2027. Until then, the shared token does nothing at Google or Bing.
It is also a draft, not a finished standard. The IETF draft is an individual proposal with no formal standing yet, and the token's name or rules could change before it's final. Google's wording is "aims", "plans" and "intends", dates may slip, and each provider will decide how it applies the signal. Check Google's crawler documentation before relying on the token.
Frequently Asked Questions
Does letting AI train on my site help it recommend my business?
None of these announcements say. If you're unsure, staying fully open is the simplest way to make sure no AI tool is left out. If you want AI tools to recommend you, focus on what they read at question time; our guide to getting your business mentioned by ChatGPT walks through it.
Will opting out remove content AI models already learned from?
None of the announcements address models already trained. The draft ties the preference to the robots.txt in effect when a page is fetched, so treat it as forward-looking.
Do I need a developer to add the opt-out?
Not usually. It's a few lines in a plain text file at the root of your site. If your website platform lets you edit robots.txt, you can add them yourself; if it doesn't, your web designer can.