NEWS
On October 30, 2026, Google announced it plans to read, in 2027, a shared robots.txt signal telling AI developers not to train on your site. There's nothing to do today, and your presence in Google Search doesn't depend on it. But opting out has a catch, and Oeuvria helps you choose without losing visibility.
November 1, 2026On October 30, 2026, on the Google Search Central blog, John Mueller announced that Google plans to support a shared AI training control in the robots.txt file. That's the file that gives instructions to the bots visiting your site.
The idea is simple. Instead of writing an instruction for every AI developer, a site will be able to state one preference, read by all participating providers. The mechanism comes from the work of the AIPREF group at the IETF, the body that develops Internet standards.
Google is aiming for 2027, and its documentation says early 2027. Company-specific controls such as Google-Extended will keep working alongside it. Search Engine Roundtable reports that Microsoft, OpenAI and Anthropic have taken similar steps.
The draft published at the IETF the same day gives this three-line example:
Two details matter. A "User-Agent: *" block isn't enough: without a group naming this signal, no preference has been stated. And it's a preference, not a crawling rule. The text is still an individual draft with no formal standing at the IETF, so the syntax could still change.
The good news first: Google states that Google-Extended has no effect on a site's inclusion in Google Search and isn't a ranking signal. Opting out of training won't remove you from search results.
The catch is elsewhere. Google-Extended controls two things: training for Gemini models, and grounding, meaning the content given to the model to support its answers in Gemini Apps. Google writes that sites explicitly allowing Google-Extended are eligible to appear in those responses and receive clicks. It also writes that an opt-out through the shared signal, with no mention of Google-Extended, will be treated as disallowing Google-Extended.
Reading those two texts together, a blanket opt-out made by reflex could also take you out of those responses. For a business that wants to be found, that's not a small detail.
To go further on being cited by AI, read our article on links in Google's AI answers.
Read: links in Google's AI answers
No. Google is aiming for 2027 to read this signal, and its documentation says early 2027. The proposal could still change before then. For now, the right move is to check what your robots.txt already says.
According to Google, no: Google-Extended, the control the shared signal will map to, has no effect on a site's inclusion in Google Search and isn't a ranking signal. It can, however, affect your presence in Gemini Apps responses.
Not on its own. The draft says this signal expresses a preference and isn't a crawling directive. It only counts with providers that choose to read and honor it.
What we actually apply:
On the services side: "SEO, GEO & AEO optimization" includes an opportunity audit and on-page and technical SEO optimizations; "Websites + lead capture system" covers building or rebuilding sites for service and construction businesses.
See our SEO, GEO & AEO service
There's no rush before 2027, but this is a good time to find out what your robots.txt says today and what you want it to say. Book a time slot with us to talk it through.
Source: Google Search Central, "Supporting a shared AI training control for websites" (Oct. 30, 2026) (Google Search Central)
Source: Google, common crawlers documentation (Google-Extended) (Google for Developers)
Source: IETF, "draft-kelley-aipref-training-preference-signal" draft (version 00, Oct. 30, 2026) (IETF Datatracker)
Source: Search Engine Roundtable, "Google Also Adopts New AI Training Control" (Search Engine Roundtable)
NEXT STEP
Twenty minutes, no commitment: we look at what you already have online and tell you what's worth changing.
Pick a time slot