Google-Extended
Google-Extended is a robots.txt token that controls whether Google may use your pages to train Gemini models and ground Gemini answers.
Google-Extended is a name you can use in your robots.txt file to tell Google whether the pages it crawls from your site may be used for its Gemini models. Google's crawler documentation says it covers training future Gemini models and grounding, which is when Gemini uses content from Google's index to answer at the moment someone asks.
People often expect it to do more than it does, so it's worth knowing exactly what it covers.
How it works
Google-Extended has no crawler of its own. You won't see it in your server logs. Google crawls your site with its usual bots, and the token only tells Google how it may use what it collected.
To opt out, you add a group for it in robots.txt:
User-agent: Google-Extended
Disallow: /What it doesn't cover
Google says Google-Extended doesn't affect whether your site is included in Google Search, and isn't used as a ranking signal. It also isn't the control for AI Overviews and AI Mode. Those are part of Google Search, and Google offers a separate setting in Search Console for them, the Search generative AI control.
For example, say Kestrel Roasters adds the rule above. Its pages stay in Google's results, and AI Overviews can still use them. What changes is Google's permission to use Kestrel's content for Gemini models and to ground Gemini's answers.
Should you block it?
It depends on what you want from Gemini. Blocking it keeps your content out of training, but it may also keep your pages out of what Gemini can draw on when it answers. If you want Gemini to describe you from your own pages, think twice.
In AnswerOrbit, the Audit tab in Website reads your robots.txt every day and keeps every version in History, with the lines added and removed each time, so you can see when a Google-Extended rule was added or changed.