No. Blocking Google-Extended in your robots.txt file does not take your site out of AI Overviews. Google’s own crawler documentation says that token does not affect whether you show up in Google Search, and AI Overviews are a Search feature.
What the Google-Extended line actually turns off
Google’s crawler documentation on Google Search Central was last updated July 14, 2026. It calls Google-Extended “a standalone product token that web publishers can use to manage whether content Google crawls from their sites may be used for training future generations of Gemini models that power Gemini Apps and Vertex AI API for Gemini and for grounding … in Gemini Apps and Grounding with Google Search on Vertex AI.”
Plain version: that line is about the Gemini app and Google’s tools for building AI products. It is not about the answer box at the top of a Google results page.
The same page says it in one sentence: “Google-Extended does not impact a site’s inclusion in Google Search nor is it used as a ranking signal in Google Search.”
There is also no bot by that name to hunt for in your server logs. Google says Google-Extended “doesn’t have a separate HTTP request user agent string” and that the token is used “in a control capacity” only.
Why AI Overviews sit on the other side of that fence
AI Overviews run off Search crawling. Google’s crawler page says preferences addressed to Googlebot “affect Google Search (including Discover and all Google Search features)”. Its page on AI features and your website, last updated December 10, 2025, says it from the other direction: “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 to how their sites are crawled for Search.”
That page also sets the bar for being named in an AI answer. To be shown “as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet”. And, in Google’s words, “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.”
So the only robots.txt lever that reaches AI Overviews is Googlebot itself. Pull that one and you are out of the ordinary blue links too. For a plumber or a roofer, that means giving up the phone calls to avoid the answer box.
The settings that do reach AI answers
Google’s robots meta tag documentation, updated March 24, 2026, lists three controls that touch AI answers without taking you out of Search:
- nosnippet. No text snippet for that page. Google says it applies to “web search, Google Images, Discover, AI Overviews, AI Mode” and “will also prevent the content from being used as a direct input for AI Overviews and AI Mode”.
- max-snippet. Caps your snippet at a set number of characters, and Google says it “will also limit how much of the content may be used as a direct input for AI Overviews and AI Mode”.
- data-nosnippet. Marks one part of a page, on a span, div or section element, so that part alone is left out.
Read the trade before you touch any of them. nosnippet also removes the grey line of description under your blue link in normal results. On a trade site, that line is often the bit that says you do emergency callouts.
There is now a separate switch inside Search Console, under Settings, called the Search generative AI control. Google’s help page for it says the control “isn’t used as a ranking or inclusion signal affecting other parts of Search”, and that it is going out “to a subset of website owners”. That page carries no update date. That switch is a bigger topic and is covered on its own page.
Two limits to know before you edit that file
Type your own address with /robots.txt on the end into your phone browser. Whatever comes back is what the crawlers read.
Two figures from Google’s robots.txt documentation, updated July 8, 2026, are worth knowing before you edit anything. Google reads at most 500 kibibytes of that file, and “content which is after the maximum file size is ignored”. Google also “generally caches the contents of robots.txt file for up to 24 hours”. An edit is not instant. Give it a day before you decide nothing happened.
While the file is open, look for “Disallow: /” under “User-agent: Googlebot” or “User-agent: *”. That is the line that can genuinely cost you work. A Google-Extended block sitting above it is the smaller worry by a mile.
Where the evidence stops
All of the above is Google describing Google. It says nothing about anyone else. OpenAI, Anthropic and Perplexity run their own crawlers under their own names, and a Google-Extended line does nothing to any of them. Whether your site is blocked from ChatGPT covers that side.
Google’s page also does not say what happens to content already used for training before you added the line. It describes what the token manages going forward, and stops there. Anyone who tells you a robots.txt edit pulls your words back out of a model is telling you something Google has not published.
Robots.txt is a request, not a lock. It is a text file asking crawlers to behave, and well run companies do. If your reason for blocking AI training is really about rights, contracts or a licensing deal, a text file is not what settles that. This is not legal advice or tax advice. Who may use your words and your job photos is set state by state, so ask an attorney licensed in your own state before you treat a robots.txt line as protection.
One more honest bit. Nobody should sell you this as a service. Adding or removing one line in a text file is a five minute job for whoever holds your site. If an agency quotes you a monthly fee for AI protection and it turns out to be a robots.txt edit, you are paying for typing, and the monthly list you should be getting instead is a good deal longer than that.
What to do next
- Open your site address with /robots.txt on the end and read the whole file. It is usually short.
- If Googlebot is blocked anywhere, fix that first. That is a Search problem, not an AI problem.
- Decide about Google-Extended on purpose. Left open, your pages may be used for grounding answers in the Gemini apps. Blocked, your Search listing is unchanged either way.
- If you want one part of a page kept out of snippets, such as a price you would rather not see quoted, use data-nosnippet on that part instead of nosnippet on the whole page.
- Make one change at a time, then wait a day and look again.
If you would rather see what the machines already say about you, the free scan asks AI assistants the questions your customers ask and writes down who gets named. The search checks page explains what gets done month to month. Neither one promises you a mention in anyone’s AI answer, and nobody honest can.