AI opt-out (Google Search)

What Google added

Google announced a control that takes a site out of its AI features and nothing else on 3 June 2026, and began testing it with a subset of website owners in the UK. It now sits in Search Console for every site, and the help page opens with a dated note:

"As of August 31, 2026, we’ve rolled out this control to all websites worldwide."

The setting is called Search generative AI and sits under Settings in Search Console. It covers AI Overviews, AI Mode and generative AI features in Google Discover, a list Google says it expects to update over time.

There are three choices. Include is the default for every property. Exclude is described like this:

"Your site's content is prevented from being visible to users in Search generative AI features, including being linked to within these features and helping with grounding AI responses in these features. You won't receive any traffic or impressions from these features."

The third choice, inherit, applies when a property sits under a parent property, such as a folder under a domain. The property follows its parent until someone sets it directly.

What the page-level rules still do

The robots rules have not changed, and they work at a different level: page by page, and on more than the AI surfaces. On nosnippet, Google's robots documentation says:

"Do not show a text snippet or video preview in the search results for this page. […] This applies to all forms of search results (at Google: 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."

So nosnippet still keeps a page out of AI answers as a direct input, and it still removes that page's ordinary search snippet in the same move. max-snippet limits "how much of the content may be used as a direct input for AI Overviews and AI Mode". data-nosnippet works on a section of a page, and the documentation attaches no AI clause to it.

The difference from the Search Console setting is scope. A robots rule acts on one page and on every kind of result. The setting acts on a whole property and only on the generative AI features.

One exception sits in the robots documentation: its limitations do not affect the use of schema.org structured data, apart from description values, and "structured data remains usable for search results when declared within a data-nosnippet element".

What excluding changes, and what it leaves

The help page describes the effect in four parts.

Your content stops being an input. "Content crawled from your site won't be eligible to be used as an input to generate an AI response or preview in Search generative AI features."

The answer can look much the same. "Content from other sites will still be available in those features, and it may appear similar to yours." Excluding a site takes out its links and its content, not the topic.

The rest of Search is not the target. The control "isn't used as a ranking or inclusion signal affecting other parts of Search". Google adds that a site's content "may still be used to power Google Search as a whole, like helping our systems better understand the language of users’ queries and the language on pages on the web."

It takes days. "Content will be excluded within 1-2 days after the control goes live, but some content may take longer to be excluded due to caching and propagation across Google systems."

Inheritance decides which setting applies where properties overlap. A property follows its closest parent that has been set manually, and if none has, the top-level domain property. A URL-prefix property for one folder can therefore differ from its domain, provided its owner sets it there.

Google-Extended is about training

Google-Extended is still the control people reach for first, and it answers a different question. The Search Console help page says so:

"This control doesn't affect AI training; to limit training of the models used to generate responses in Search generative AI features, use Google-Extended."

The crawler documentation describes the token as managing whether crawled content "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 on Vertex AI. It adds:

"Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search."

Put side by side, the Search Console setting decides whether a site appears in and feeds the AI answers in Search, and Google-Extended is the control Google names for training. Setting one does not set the other. And Google-Extended does not show up in server logs as a visitor of its own: "Google-Extended doesn't have a separate HTTP request user agent string."

Before you switch it off

Measure first. Google points to the generative AI performance report in Search Console, which shows impressions in AI Overviews and AI Mode by page, country, date and device, and says it "can help you get an idea of how changing your control may impact traffic to your site." After excluding, there is nothing left to measure.

Check which property you are changing. A setting on a domain property flows down to every folder and subdomain property below it, unless one of them is set on its own.

Keep nosnippet for page-level decisions. It still costs the page its ordinary snippet in web search.

Decide training separately. If the concern is model training rather than appearance, the instrument is Google-Extended, and the Search Console setting does not do that job.