On October 30, 2026, Google and Microsoft both said they will honor a single line in a website’s robots.txt file that tells AI companies not to train their models on the site. Search Engine Roundtable reported that OpenAI and Anthropic are backing the same signal. For a publisher, this is a simpler way to say no. For a plumber, dentist or law firm, the bigger risk is saying no by accident, because some of these controls also decide whether your site can feed the AI answers your customers read.
What Google and Microsoft announced
-
One shared opt-out. In a post on the Google Search Central blog, John Mueller, Google Search Advocate, wrote that Google plans to support “a shared AI training control in robots.txt.” He said it “will help websites express their AI training preferences across multiple AI providers in a single step.”
-
The line itself. Google’s crawler documentation now says Google “intends to support the AIPREF shared machine-readable training preference signal (with “user-agent: __usage_ai-training”) in early 2027.” The IETF draft gives this example of a site opting out everywhere except one folder:
User-Agent: __usage_ai-training Disallow: / Allow: /press-kit/ -
How Google will read it. Mueller wrote: “When a website uses this policy token to opt out of training in its robots.txt file (and Google-Extended is not explicitly specified), Google plans to interpret that as if the publisher had disallowed Google-Extended.” Google-Extended is the AI training control Google has offered since 2023. Existing controls like it “will continue to function alongside this new joint signal.”
-
It is still a draft. The proposal is an Internet-Draft written by Kevin Kelley of Anthropic, dated October 30, 2026. The IETF page says it “is not endorsed by the IETF and has no formal standing in the IETF standards process.” Google says it aims to support the control “in 2027.”
-
Microsoft goes further. In a post by Krishna Madhavan, Principal Product Manager, Microsoft Bing, Microsoft said it will honor the shared training line and add a separate “AI Grounding” preference set page by page. Grounding means using your content to inform an AI answer when it is written. Microsoft says “Training and grounding are independent preferences,” and that it will start honoring the new settings on February 1, 2027.
What it means for a local service business
A local business has a different goal from a newspaper. A newspaper may lose money when an AI summarizes its articles. A roofer or an HVAC company makes money when someone calls. Most of the time you want AI tools to know who you are, where you work and what you do.
Training and AI answers are not the same switch
Microsoft keeps them apart. You can block training and still allow grounding, and Microsoft says grounding settings “do not change eligibility for traditional Bing search results.” It is also blunt about the cost of blocking grounding: “If you select DisallowExtended or Disallow for grounding, you will see reduced brand visibility in AI experiences.”
Google’s setup is less separated. Its crawler page says Google-Extended covers both training for Gemini models and grounding in Gemini Apps. Mueller’s post says sites “that specifically allow Google-Extended are eligible to appear in responses and receive clicks via grounding in Gemini Apps.” So when Google reads the shared opt-out as a Google-Extended block, the effect may reach past training into Gemini Apps answers. Google’s page also says Google-Extended “does not impact a site’s inclusion in Google Search nor is it used as a ranking signal in Google Search.”
Look at your robots.txt now
Open yoursite.com/robots.txt in a browser. Look for lines that name Google-Extended, __usage_ai-training or other AI crawlers with Disallow: / under them. If a web designer, an agency or a plugin set up your site, it may include AI rules you never chose. If you find one, ask whoever added it what it was meant to do, and whether it was meant to keep you out of AI answers too.
Do not paste a “block all AI” template
If someone offers you a template that blocks AI crawlers, then before you copy it, decide what you are trying to protect. If there is one folder you do not want used, such as original photos or a paid guide, the draft lets you block a single path instead of the whole site.
If your site uses NoArchive or NoCache, note March 1, 2027
Microsoft says these older tags will stop being read as training and grounding preferences. Sites that use them and want to keep their restrictions must set the new preferences “before March 1, 2027.” Grounding preferences also need a temporary setting in Bing Webmaster Tools, which opens December 15, 2026. If your site has never used these tags, Microsoft says no transition action is needed.
What did not change
Nothing takes effect today. Google says 2027, and Microsoft starts on February 1, 2027.
The new line does not block crawling. The IETF draft says “This policy token is not a directive for crawling behavior.” It also says a User-Agent: * group “MUST NOT be interpreted as matching the policy token,” so your existing catch-all rules will not count as an opt-out.
Google AI Overviews and AI Mode have their own control in Search Console, which we covered in our post on the Search Console AI setting. This announcement does not change it.
Allowing everything is not a promise either. Microsoft says: “Neither choice guarantees inclusion, traffic, or payment.” No AI engine can be bought or guaranteed. A fixed set of buyer questions re-run every month is still the way to see whether AI answers actually name your business.
The short version
Google and Microsoft will honor a shared robots.txt line, __usage_ai-training, that opts a website out of AI training, starting in 2027. Google will read it as a Google-Extended block, and Google-Extended also covers grounding in Gemini Apps. Most local service businesses want to be found in AI answers, so check your robots.txt for AI blocks you did not ask for, and do not add one without knowing what it switches off.