Does blocking Google-Extended remove me from AI Overviews?

No. Google's crawler documentation 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. AI Overviews are a Search feature built on the Search index that Googlebot fills, so the robots.txt control for them is Googlebot.

What does Google-Extended actually control?

Whether content Google already crawled may be used to train future Gemini models, and whether it is used for grounding in Gemini Apps and Grounding with Google Search on Vertex AI. Google states it is a control token, not a crawler: it has no separate HTTP user agent string of its own.

How do I limit what AI Overviews quote from my pages?

With the snippet controls: nosnippet, data-nosnippet, max-snippet, or noindex. Google names these in its AI features guidance. All of them also restrict your ordinary Search snippets, which is the tradeoff you are accepting.

Is there any cost to blocking Google-Extended?

Yes, just not in Search. You give up being used for Gemini model training and for grounding in Gemini Apps and Vertex AI, so you are less likely to be a source Gemini pulls from at prompt time. Whether that matters depends on whether Gemini sends you anything.

Blocking Google-Extended does not affect AI Overviews

Blocking Google-Extended does not remove you from AI Overviews. Google says so in one sentence on its own crawler page: "Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search" (Google Search Central, page last updated 2026-07-14).

AI Overviews are a Search feature. They are assembled from the Search index that Googlebot fills. The robots.txt token that governs that index is Googlebot, and the only way to keep yourself out of AI Overviews with robots.txt is to keep yourself out of Google entirely.

The short answer

Googlebot governs the Search index, and AI Overviews are served from it.

Google-Extended governs Gemini training and Gemini grounding. It is not a crawler at all.

Blocking Google-Extended costs you nothing in Search or AI Overviews, and costs you real presence in Gemini Apps.

The only levers that limit AI Overviews are the snippet controls, and they cost you ordinary snippets too.

Google-Extended is a control token, not a crawler

Google-Extended never fetches anything. Its documentation says so directly: "Google-Extended doesn't have a separate HTTP request user agent string. Crawling is done with existing Google user agent strings; the robots.txt user-agent token is used in a control capacity."

Googlebot fetches the page. Google-Extended decides what happens to the copy afterwards. If you go looking for Google-Extended in your access log you will never find it, because it does not exist as traffic.

What it decides is spelled out on the same page. In Google's wording, the token manages whether content "may be used for training future generations of Gemini models that power Gemini Apps and Vertex AI API for Gemini". It also covers "grounding (providing content from the Google Search index to the model at prompt time to improve factuality and relevancy) in Gemini Apps and Grounding with Google Search on Vertex AI."

Two jobs, then: training future Gemini models, and grounding today's. Calling this a pure training opt-out drops the grounding half. Blocking Google-Extended means Gemini is less likely to pull your page in at prompt time to answer a question, which is a retrieval consequence today rather than a hypothetical about a model two years out.

Where AI Overviews actually get their sources

From Search. Google's AI features guidance puts it flatly: "AI is built into Search and integral to how Search functions, which is why robots.txt directives for Googlebot is the control for site owners to manage access" (Google Search Central, retrieved 2026-08-17).

The same page says there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." There is no separate crawler, no separate schema, and no separate opt-in. If Googlebot can read the page and the page is indexed, it is eligible.

Google does link to Google-Extended from that page. The wording puts it outside Search: "To limit AI training and grounding in some of Google's other systems, read more about Google-Extended."

What each Google token controls

Token	Product it affects	Effect on AI Overviews
`Googlebot`	The Google Search index	Total. Block it and you leave Search, AI Overviews included.
`Google-Extended`	Gemini model training, grounding in Gemini Apps and Vertex AI	None. Google states it is not a Search inclusion or ranking signal.
`GoogleOther`	Nothing specific. Google calls it "the generic crawler that may be used by various product teams" for things like "one-off crawls for internal research and development"	None.
`Google-CloudVertexBot`	Crawls that site owners request for building Vertex AI Agents. Google: "It has no effect on Google Search or other products"	None.
`Google-InspectionTool`	The Search Console URL Inspection and Rich Results tests	None, but blocking it breaks your own debugging tools.

Every row except the first is safe to block from a Search perspective and pointless to block if AI Overviews were your reason.

The controls that do limit AI Overviews

There are exactly four, and Google names them together: nosnippet, data-nosnippet, max-snippet, and noindex.

They work because AI Overviews quote from the same indexed content that feeds a normal Search snippet. Restrict the snippet and you restrict what the overview can lift. That symmetry is also the catch:

nosnippet removes your snippet from ordinary Search results as well. You keep the blue link and lose the description under it.
data-nosnippet is the surgical one. Wrap a specific element and only that element is excluded, so you can protect a pricing table or a proprietary paragraph without going dark on the whole page.
max-snippet:[n] caps the character count. Useful when the goal is "quote less of me", not "quote none of me".
noindex removes the page from Search entirely. It works, and it is almost never what a publisher wants.

If a page is worth ranking, data-nosnippet around the part you do not want quoted is usually the only one of these worth deploying. The other three trade Search visibility for AI Overview control at roughly one to one.

The Google fetches your robots.txt does not govern

Google keeps a third list separate from its crawlers, and it changes what a robots.txt audit can promise. Its user-triggered fetchers page states: "Because the fetch was requested by a user, these fetchers generally ignore robots.txt rules" (Google Search Central, retrieved 2026-08-17).

Google-GeminiNotebook is on that list. So is Feedfetcher, and the Chrome Web Store fetcher. These are requests a person set in motion, and Google treats them the way OpenAI treats ChatGPT-User: as an action by a user, not a crawl.

The practical consequence is that "I blocked Google-Extended so Gemini cannot see my content" was never true even before you read the Search caveat. A person can paste your URL into a Gemini surface and get your page. What robots.txt controls is systematic collection, not individual retrieval, and every major operator now draws that line the same way.

If you need a fetch actually refused rather than politely discouraged, that is a job for authentication or an IP rule at the edge, not for a text file that asks nicely.

What we block, and what it costs us

Our production robots.txt used to disallow Google-Extended. It arrived there through Cloudflare's Managed robots.txt feature, which injects a fixed list of AI crawler tokens into the served file at the edge. Nothing in src/app/robots.ts mentioned Google-Extended, or any AI crawler, which is why the file on localhost and the file on the live domain were different documents.

What did that rule cost us? In AI Overviews, nothing, and AI Overviews are the surface we care about most. In Gemini Apps, it meant our pages were not eligible to be pulled in as grounding at prompt time. That is a real cost, and we had not decided to pay it, which is a different problem from paying it.

We removed it on 2026-08-15. Not because the Gemini grounding traffic was measurable, it is not, and we will not invent a number for it. We removed it because a rule that is defensible and a rule that is examined are not the same thing, and this one had never been either written or reviewed by us. The file customers read is now the file in the repository.

What to check on your own site

Fetch the live file, not the one in your repo: curl -s https://yourdomain.com/robots.txt. A CDN, a host, or a security plugin can all rewrite it after your code runs, and ours does.

Then ask three questions in order:

Is Googlebot disallowed anywhere, including inside a broad group? That is the only robots.txt rule that removes you from AI Overviews, and it removes you from Search with it.
Is Google-Extended disallowed, and did anyone decide that? If the reason was AI Overviews, the rule is not doing what you think and you can drop it.
Is there a nosnippet on pages you want ranked? Check the page HTML and your X-Robots-Tag headers, not just robots.txt.

Our AI crawler checker runs the first two against a live robots.txt. It prints the operator's own wording next to each verdict, so you can read what a token controls instead of trusting a summary. It is free and takes no account.

One disclosure since we sell here: SEOBuilder asks seven answer engines the same buyer questions and reports the citation URLs each one showed. Gemini, ChatGPT, Perplexity and Copilot are read as the products themselves, and Claude answers through Anthropic's API rather than claude.ai, which we would rather spell out than let a logo imply the consumer app.

Once the tokens are right, the next question is which pages get cited and why, which is a ranking and content problem rather than a crawler one. How Google AI Overviews choose citations covers what shows up in the citation list, how to measure AI visibility covers how to count it without guessing, and tracking AI visibility in Google AI Overviews is the same job done for you.