Google-Extended: what it blocks and what it does not

Blocking Google-Extended does not remove you from Google Search. It is a robots.txt token that lets publishers manage whether content Google crawls may be used for training future generations of Gemini models, and Google's documentation says it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search". It is a separate control from the one that governs Googlebot.

Quick version

• To opt out: add a group with User-agent: Google-Extended and Disallow: /. Search Engine Journal quotes that syntax from Google's documentation.
• What it covers: training of Gemini models, and grounding. PPC Land reports that as of September 2026 the documentation defines those two uses.
• What it does not cover: Search inclusion or ranking, per Google's crawler page.
• What you will not see: any trace in your server logs. More on that below.
• Related: rules for ChatGPT are set separately. See our guide to blocking ChatGPT in robots.txt.

What Google-Extended is

Google-Extended is a control token, not a bot. Google says it doesn't have a separate HTTP request user agent string. PPC Land puts it more bluntly: no request ever arrives from Google-Extended, and no server log will record one.

Crawl Lab, a site that publishes crawler guides, checked its own logs. It reports that across 1,722 crawler requests analysed over eight days, Google-Extended does not appear once. Its explanation is that Googlebot and other Google crawlers collect content once, then Google-Extended acts as a downstream usage switch.

The token has changed over time. Search Engine Journal reports it was introduced on September 28, 2023, to manage whether sites help improve Bard and Vertex AI generative APIs, and that Bard was later renamed Gemini Apps. Google's changelog says that based on publisher feedback it updated the token description for more specificity and clarity. As of September 2026, the documentation defines two uses: training future generations of Gemini models, and grounding.

Grounding is the less obvious one. Capconvert, a company that sells AEO audits and optimization services, describes it this way: when Gemini pulls current information from the Search index and attributes it, that is grounding at work. Treat that as one vendor's explanation and check Google's own page for the current wording.

What blocking it does and does not touch

A few points behind the table.

Search is driven by Googlebot

Crawl Lab says that blocking Google-Extended changes nothing in Google Search: indexing, ranking and AI Overviews inclusion remain driven by Googlebot. That is why blocking this token and blocking Googlebot are different decisions with different consequences. Google's crawling infrastructure is also shared across a variety of Google products beyond Search, so a rule aimed at one product does not automatically apply to another.

It does not undo past use

Capconvert says the directive is forward-looking: it governs future training and current grounding, and it does not retract content already used in a shipped model. PPC Land adds that Google told the UK Competition and Markets Authority that removing content already added to its search AI models was not feasible. If your goal is to pull back content already used, a robots.txt line will not do it.

It is robots.txt only

Capconvert states that Google-Extended honours only robots.txt, with no meta tag equivalent, so page-level control is not available. You set the rule for your site, or for the paths you list.

Other Google tokens do affect Search features

Do not copy a Google-Extended rule onto other Google agents without reading what they do. Per Google's crawler documentation:

• Googlebot-Image preferences affect Google Images, Discover, Video mode, and all features in Google Search where images, logos, and favicons are presented.
• Storebot-Google preferences affect all surfaces of Google Shopping.
• Google-InspectionTool preferences affect Search testing tools such as the Rich Result Test and URL inspection in Search Console.
• GoogleOther is a generic crawler that may be used for one-off crawls for internal research and development.

Google's changelog also notes that Mediapartners-Google preferences affect multiple ad-related products, including AdSense and Ad Manager. Blocking any of these is a different choice from blocking Google-Extended.

What blocking may cost you

The cost is not in Search rankings, per Google. It is in the AI side, and the evidence there is thin and mixed.

Capconvert argues that if you disallow Google-Extended, you are opting out of being the grounded source, which means opting out of the citation that comes with it. That is a vendor's reading, and the company sells AEO services.

Two data points from PPC Land pull in different directions:

• BuzzStream found that 92.3% of top news sites blocking Google-Extended still appeared in AI citations across ChatGPT, Gemini, AI Overviews and AI Mode.
• Ron Berman of Wharton found that news publishers blocking large language model crawlers, a set that included Google-Extended, lost about 7% of weekly visits within six weeks.

Both studies are about news publishers, and the second covers a group of crawlers rather than Google-Extended alone. Neither tells you what would happen on a small business site. Nothing here promises that blocking or allowing will change your citations or traffic.

Our view: this is a content-licensing decision, not an SEO fix. If you do not want your pages used to train Gemini and you accept the possible loss of grounding, block it. If you want to be quoted by AI tools, leaving it open is the more consistent choice. This is our judgment, not a sourced fact.

How to add the rule

1. Open your robots.txt. It sits at the root of your domain.
2. Add the group. Per Search Engine Journal, the syntax is:

User-agent: Google-Extended
Disallow: /

3. Leave your Googlebot rules alone. Do not paste Disallow: / under a Googlebot group. That is a different instruction.
4. Mind the group rules. PPC Land says Google's parser follows only the most specific group matching a user agent. Our reading, not a sourced fact: do not assume rules from your general * group carry over to a Google-Extended group.
5. Wait for the cache. The same source says Google generally caches robots.txt for up to 24 hours.
6. Verify the file, not the logs. Open your live robots.txt and read it. Your logs will not help, because you will never see Google-Extended in server logs. Crawl Lab adds that any user agent presenting itself that way is an impostor.

Google says its common crawlers always obey robots.txt rules when crawling automatically, so the file is the right place for the rule.

If you want a quick look at how AI tools see your site before you decide, our free AI check is one place to start.

Check the current wording before you act

This area moves. Google's changelog shows the token description was revised, and the crawling docs have migrated to a new documentation site. PPC Land reports that the CMA designated Google with strategic market status in general search on October 10, 2025 and imposed a Publisher Conduct Requirement on June 3, 2026. That is context, not advice on what it requires of you. Read Google's common crawlers page for the current wording on the day you edit your file.