Skip to content
Adrythm
Book 20 minutes
AI search

Google-Extended

blocking AI from your content / AI crawler control / data-nosnippet / opting out of AI training

In short

Google-Extended is a robots.txt token that controls whether content Google crawls from your site may be used to train Gemini models and ground Gemini apps. Google states it does not affect your inclusion in Search or your ranking. It is a training control, not a way to stay out of search results.

The mechanics are unusual and worth knowing first. Google-Extended has no user agent string of its own. Google says crawling is done with existing Google user agent strings, and the robots.txt token is used in a control capacity. So nothing new visits your site. You are adjusting what an existing visit is allowed to feed.

What it governs is narrow and specific. Google describes it as a product token publishers use to manage whether crawled content may be used for training future generations of Gemini models, and for grounding, which Google defines in the same sentence as providing content from the Google Search index to the model at prompt time to improve factuality and relevancy. The products named are Gemini apps and Vertex AI. Those are the ones the documentation covers.

Now the sentence that settles most arguments about it. Google states that Google-Extended does not impact a site's inclusion in Google Search, nor is it used as a ranking signal in Google Search. Blocking it is not a way to disappear, and it is not a way to be rewarded either.

That leaves a second question people usually have bundled with the first, and it needs a different tool. Bing added support for the data-nosnippet HTML attribute in October 2025, which marks sections of a page so they do not appear in snippets or AI-generated answers. Bing is explicit about the trade: content marked that way is still indexed normally, is still available for ranking, and is simply excluded from snippets and AI summaries.

Google's own lever for the summaries is a third thing, and it is blunt. The nosnippet rule takes your text snippet out of every kind of Google result, and Google says it will also prevent the content being used as a direct input for AI Overviews and AI Mode. There is no setting that keeps the snippet and refuses the summary. Google adds that where robots rules conflict the more restrictive one wins, so a page carrying both max-snippet:50 and nosnippet ends up with nosnippet.

In practice

Someone asks whether the business can stop AI using its content. There are two questions hiding in that, and they have different answers. Do you object to your material training a model, or do you object to it being quoted back inside an answer? The first is a robots.txt decision. The second is a markup decision, made per section, and a robots.txt line will not do it.

Not the same as

Disallowing a crawler
That stops the content being fetched at all. This permits the fetch and limits one downstream use.
noindex
That removes the page from results. Both controls here leave the page indexed.

Why it matters to you

This is usually presented as a single switch labeled block AI, and it is not one. The controls sit at different layers, are set in different places, and are honored by different engines. Before changing either, it is worth knowing which outcome is actually wanted, because the documentation for these two levers describes different things: one is about training and grounding, the other is about what appears in a snippet or a summary.

What to ask or check

  1. 01Which are we trying to control, what trains a model or what gets quoted?
  2. 02If we block this, what are we giving up, and who decided that trade is worth making?
  3. 03Are we applying the same decision to other engines, which use their own controls?

What people get wrong

That it is one switch that keeps your content out of AI answers. It is a token governing training and grounding for the products Google names, and Google says it is not a ranking signal and does not affect inclusion in Search.

Red flags

  • A proposal to block AI that only edits robots.txt and never mentions snippet controls.
  • A claim that setting this will protect or improve rankings, when Google states it is not a ranking signal.
  • One engine's control assumed to cover every engine.

Who owns it

Whoever can edit robots.txt sets the first lever. The second is markup, so it belongs to whoever controls the page templates.

Where you will see it

In robots.txt, and in proposals about protecting content from AI.

Keep reading

AI Overviews

An AI Overview is the summary Google sometimes places above the results, with links to sources. Google says they are only shown when its systems judge them additive to classic Search, so they often do not trigger. There is no special file or markup that gets you into them.

robots.txt

A robots.txt file tells crawlers which URLs they may fetch on your site. Google is explicit that it is not a way to keep a page out of search: a disallowed URL can still be indexed if other sites link to it. The convention dates to 1994 and became RFC 9309 in 2022.

AI performance report

An AI performance report shows how often AI search features show or cite your site. Google's version counts impressions in AI Overviews and AI Mode. Bing's counts citations and states its numbers do not indicate ranking, authority or placement. Neither publishes a position inside an AI answer.

AI content disclosure

AI content disclosure is labeling material as machine-made where a platform requires it. Google's product listings require AI-generated images to carry a specific IPTC metadata code, and IPTC defines three different codes: created with generative AI, edited with it, and edited by a person.

Grounding

Grounding is tying an AI answer to retrieved sources rather than letting the model answer from memory. Google uses the word as a synonym for retrieval augmented generation. Microsoft treats groundedness as something you measure on a scale, which is the more useful way to think about it.

Query fan-out

Query fan-out is the model turning one question into several searches at once. Google describes it as a set of concurrent, related queries generated by the model to fetch additional relevant search results. The trap is the obvious response: Google names a page per fan-out query as scaled content abuse.

Want this explained against your own numbers?

Twenty minutes, a straight answer, and no follow-up sequence if you decide not to work with us.

Book 20 minutes