Google's AI search opt-out switch: where it is and what it does

Google's opt-out switch for AI search lives in Search Console, under Settings > Search generative AI. It lets you include or exclude your whole site (or one part of it) from AI Overviews, AI Mode and the generative AI features in Google Discover. Flip it to exclude and your links and content leave those features, usually within a few days.

Here is the part most people miss: the switch is a trade, not a free privacy setting. In its announcement, Google says plainly: "Sites that opt out will not receive traffic or impressions from our generative AI features." So before you touch it, you want to know what those features send you today.

By the end of this guide you will be able to find the control, pick the right option for your site, measure what you would lose first, and use page-level tags when you only want to hide a few pages instead of the whole site.

What the Search generative AI control actually is

It is a single setting per Search Console property. A property is the site, or part of a site, that you have proved you own in Search Console, such as example.com or https://example.com/blog/. At the time of writing (26 October 2026), Google's help page lists three features it covers: AI Overviews, AI Mode, and generative AI features in Google Discover. Google says it expects to update that list over time.

Think of it like the sign on your shop door that says "no flyers". You are not closing the shop. You are telling one kind of visitor, the one who summarises your shop for other people, not to quote you. The normal Google results, the blue links, keep working.

Google announced the control on 3 June 2026, first for a subset of site owners in the UK. The blog post was updated on 31 August 2026 with this line: "As of August 31, 2026, we've rolled out these features to all websites worldwide."

What it does not do

This is where people get confused, so let me be very direct. Google's help page lists these limits:

• It does not affect AI training. For training, Google points you to a separate control called Google-Extended. I cover that in a separate post on robots.txt rules for GPTBot, ClaudeBot and Google-Extended.
• It does not remove you from Google Search. To block a page from Search completely, Google says to use noindex.
• It does not override your choices in other services, such as Merchant Center and Google Ads.
• It is not a ranking signal outside the AI features. The announcement says: "This control will not be used as a ranking signal for search results outside of these generative AI Search features."
• Your content may still be used to power Google Search as a whole, for example to help Google understand the language of queries and pages.

Why this matters: if your worry is AI training, this switch is the wrong tool. If your worry is AI summaries answering for you, it is the right one.

Before you start

You need very little, but check each item before you change anything.

• A verified Search Console property. If you have not verified your site yet, follow my guide to verifying a domain in Google Search Console with a TXT record first.
• Owner access. Google's permissions page says owners "can add and remove other users, configure settings, view all data, and use all tools." If you are a full or restricted user, ask an owner to make the change or give you owner rights.
• Knowing your property tree. If you have a domain property like example.com plus URL-prefix properties like https://example.com/blog/, note which one you are changing. Settings flow down from parent to child.
• Some baseline numbers. You will check the Generative AI performance report before you exclude. Once excluded, that data may stop showing.
• For page-level tags only: access to your page templates or your server config (for example .htaccess on Apache), plus a backup copy of any file you edit.
• Time: about 10 minutes for the switch, plus a few days for Google to apply it.
• Risk level: low for the switch, because it is reversible. Medium for server config edits, because a typo in .htaccess can take your site down until you fix it.

How to find and change the switch

Step 1: Open the right property in Search Console

Sign in to Search Console and pick the property you want to change. If you manage several properties for the same site, start with the top-level domain property, because children inherit from it by default.

Quick tip: Google's help page says the control for a property follows "its closest parent that has changed its control to stop inheriting." If no parent has done that, it follows the top-level domain property. So one change at the top can cover the whole site.

Step 2: Check what AI search sends you today

Do this before you change anything. Open the Generative AI performance report. The control help page links to it directly, and it sits with the other performance reports in Search Console.

Google defines impressions in this report as "how many times links to your site were shown to a user in a generative AI feature on Google Search." It covers AI Overviews and AI Mode. Google Discover has its own separate generative AI performance report.

You can group the data by these tabs:

• Pages: the final URL linked, after redirects, mostly assigned to the canonical URL.
• Countries: where the search came from.
• Dates: by day, week or month. All dates are in Pacific Time (PT).
• Devices: desktop, tablet or mobile.

What you should see: a chart of total impressions and a table below it. The table has the usual 1,000 row limit of the Search performance report, so use the export button if you want to save a copy.

Heads up: if the report does not appear at all, Google says this "usually happens when your site hasn't received enough impressions in generative AI features on Google Search." It can also mean the site is already excluded. Either way, that tells you something useful.

Made-up example: Sara runs a small recipe site. She checks the report and sees her top 20 pages get most of their AI impressions from mobile users in two countries. Instead of excluding the whole site, she decides to protect only her paid meal-plan pages with page-level tags (covered below) and leave the free recipes included. Sara and her numbers are invented for this guide, but the decision process is the point.

Step 3: Open Settings and the Search generative AI control

Open Settings by clicking the settings icon in the navigation pane. Google's help page says: "See and manage the Search generative AI control for your property under Settings > Search generative AI in Search Console."

What you should see: the Search generative AI control page with three options, one of them already selected.

Step 4: Choose one of the three options

Google's help page gives these exact labels:

1. Include my site's links and content in Search generative AI features. This is "the default control for all properties." Your content can show as links and help ground AI answers, and you can get impressions and traffic.
2. Exclude my site's links and content from Search generative AI features. Your content is kept out of these features, including links and grounding. "You won't receive any traffic or impressions from these features."
3. Inherit control from parent. Your property follows its parent's value. "This is the default if your property has a parent property."

Grounding, in plain words, means Google uses your page as a fact source when it writes an AI answer. If you exclude, Google says content crawled from your site "won't be eligible to be used as an input to generate an AI response or preview" in these features.

Heads up: excluding your site does not make the topic disappear from AI answers. Google says content from other sites "may appear similar to yours." If a competitor writes about the same thing, the AI can still answer from them.

Step 5: Set a different choice for one section (optional)

You might want the blog included but a members area excluded. To do that, you need a separate property for that section, such as the URL-prefix property https://example.com/members/. Google says to "log in to that property in Search Console and update its configuration using the Search generative AI control page."

Owners of the top-level domain property "will see a notification on the Search generative AI control page when one of their child properties overrides the parent setting." So if you are the main owner and a colleague changes a child, you will hear about it.

Step 6: Wait, then measure

Google says it "generally takes a few days" for content to be excluded. The page also says 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."

To undo, go back to the same control page and pick Include or Inherit control from parent.

Page-level controls: when you only want to hide some pages

The site switch is all or nothing for a property. If you only want to limit a few pages, Google's documentation points to robots meta tags, which are small instructions in a page's HTML, and the X-Robots-Tag HTTP header, which sends the same instructions in the server response.

Why this matters: these tags also change your normal search snippets, the short text under your blue link. So they cost you something in regular Search too. Use them on purpose.

Option A: nosnippet on one page

Google defines nosnippet as: "Do not show a text snippet or video preview in the search results for this page." The same docs say it applies to AI Overviews and AI Mode and "will also prevent the content from being used as a direct input for AI Overviews and AI Mode."

This line goes inside the <head> of the page and tells all search crawlers not to show a snippet:

<meta name="robots" content="nosnippet">

Option B: max-snippet to limit length

The max-snippet rule means "Use a maximum of [number] characters as a textual snippet for this search result." Google says it also limits "how much of the content may be used as a direct input for AI Overviews and AI Mode." A value of 0 is equivalent to nosnippet.

This tag allows a snippet of up to 20 characters only:

<meta name="robots" content="max-snippet:20">

Option C: hide one paragraph with data-nosnippet

If you only want to keep one block out of snippets, Google supports the data-nosnippet attribute on span, div and section elements. Other elements are not valid for it.

This HTML keeps the second block out of snippets while the first can still be shown:

<p>This text can be shown in a snippet.</p>
<div data-nosnippet>
<p>This part stays out of snippets.</p>
</div>

Option D: X-Robots-Tag on Apache with .htaccess

This is useful for PDFs and other files that have no HTML <head>. Google's docs say "Any rule that can be used in a robots meta tag can also be specified as an X-Robots-Tag."

Before you edit, back up the file. This command makes a dated copy of .htaccess in the same folder:

cp .htaccess .htaccess.backup-$(date +%Y%m%d)

Google's example adds a header to every PDF on the site. Add this block to the .htaccess file in your site root, or to httpd.conf:

<Files ~ "\.pdf$">
Header set X-Robots-Tag "noindex, nofollow"
</Files>

Heads up: that example uses noindex, which removes those PDFs from Google Search entirely, not only from AI features. If you only want to keep them out of snippets and AI answers, use nosnippet instead. This block does that:

<Files ~ "\.pdf$">
Header set X-Robots-Tag "nosnippet"
</Files>

A note on the syntax from the Apache docs: Header comes from the mod_headers module, and set replaces any earlier header with the same name. The <Files ~ "..."> form takes a regular expression, though Apache's docs say <FilesMatch> "is preferred". This version does the same thing with <FilesMatch>:

<FilesMatch "\.pdf$">
Header set X-Robots-Tag "nosnippet"
</FilesMatch>

Option E: the same header on Nginx

Google gives a matching example for the site's .conf file on Nginx. Google's own version uses noindex, nofollow; this one uses nosnippet:

location ~* \.pdf$ {
add_header X-Robots-Tag "nosnippet";
}

The robots.txt trap

Do not block a page in robots.txt and also put nosnippet on it. Google says that if a page is disallowed from crawling, "any information about indexing or serving rules will not be found and will therefore be ignored." Google has to crawl the page to read the tag. For more on how AI crawlers and robots.txt fit together, see how to allow search crawlers while blocking AI training.

How to check it worked

Check the site switch

Go back to Settings > Search generative AI and confirm the option you picked is selected. Then, after a few days, open the Generative AI performance report. Expected result: impressions fall towards zero after an exclude, or the report stops showing, because Google notes excluded sites may not see it.

Remember the report uses Pacific Time dates, and the newest data "might change in the next few hours."

Check a header with curl -I

The curl -I command asks the server for the headers only, using an HTTP HEAD request. Run it against one of your PDFs:

curl -I https://example.com/files/guide.pdf

Expected result: somewhere in the output you see this line.

X-Robots-Tag: nosnippet

If the URL redirects (for example from HTTP to HTTPS), add -L so curl follows it and shows the headers of each page:

curl -IL http://example.com/files/guide.pdf

Check a meta tag

This command downloads the page quietly and shows only lines that mention robots:

curl -s https://example.com/members/plan/ | grep -i robots

Expected result: your <meta name="robots" content="nosnippet"> line.

Check what Googlebot saw

Google's AI features page suggests the URL Inspection tool in Search Console "to see the HTML that Googlebot received while crawling the page." Inspect the page URL and look for your tag in the crawled HTML.

Troubleshooting

Problem: you cannot see the Search generative AI option in Settings

Symptom: Settings opens, but you do not find the control, or you cannot change it. Likely cause: you are a full or restricted user, not an owner, since Google says owners are the ones who "configure settings". Fix: check Settings > Users and permissions (visible only to owners) or ask the owner to grant you owner access.

Problem: you changed the setting but nothing happened on a child property

Symptom: the blog section still shows in AI answers. Likely cause: the child property has its own manual setting, which overrides the parent. Fix: open the child property, go to its control page and set Inherit control from parent, or set it to exclude directly.

Problem: your site still appears in AI Overviews two days later

Symptom: you excluded, but you still see your link. Likely cause: Google says some content "may take longer to be excluded due to caching and propagation across Google systems." Fix: wait a few more days and check the report trend, not single searches.

Problem: the AI still answers with "your" content

Symptom: an AI Overview says almost exactly what your page says, without your link. Likely cause: Google says content from other sites may still appear and "may appear similar to yours." Fix: nothing in the switch changes this. It only controls your own site's content.

Problem: 500 Internal Server Error after editing .htaccess

Symptom: the whole site shows an internal server error. Likely cause: the Apache docs list Header with "Override: FileInfo", and the Apache docs say directives not permitted by your server's AllowOverride setting in an .htaccess file cause an internal server error. A syntax error does too, and so can a directive whose module (here mod_headers) is not loaded. Fix: restore your backup with cp .htaccess.backup-YYYYMMDD .htaccess (use the real date), then ask your host whether mod_headers is enabled and FileInfo overrides are allowed.

Problem: curl -I shows no X-Robots-Tag header

Symptom: the header line is missing. Likely cause: the <Files> pattern does not match the filename, the .htaccess file is in the wrong folder, or a CDN is serving a cached copy. Fix: check the file extension and case in the pattern, check the folder, and purge your CDN cache before testing again.

Problem: nosnippet seems ignored

Symptom: the tag is on the page but AI answers still quote it. Likely cause: the page is blocked in robots.txt, so Google never reads the tag, or Google has not recrawled yet. Fix: remove the robots.txt block for that URL, confirm the tag with URL Inspection, and allow time for Google to recrawl the page.

When to ask your host or a developer

Ask your host if you are not sure mod_headers is enabled, or if you cannot reach the Nginx config. Ask a developer if your pages come from a CMS template, since a CMS may not let you edit the page HTML or headers directly.

Security note: neither the switch nor these tags protect private content. Google's documentation says that if you have confidential or private content, you need to password protect it. If something must stay private, put it behind a login.

And if your goal is the opposite, getting cited more often in AI answers, Google says "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." A page must be indexed and eligible to show with a snippet. My post on how to get cited by ChatGPT, Claude and Google AI answers goes deeper on that side.

The one message worth forwarding

Copy this to whoever manages your site or Search Console:

"Hi, Google now has a site-wide switch for AI search in Search Console, under Settings > Search generative AI. Before we change it, can you export the last 3 months of the Generative AI performance report so we know what AI Overviews and AI Mode send us? If we exclude, we lose all traffic and impressions from those features, but normal search ranking is not affected and it does not stop AI training (that is a separate control called Google-Extended). For a few sensitive pages, we could use a nosnippet meta tag or X-Robots-Tag header instead. Can you check which option fits us and confirm the change with a curl -I test or URL Inspection?"

The takeaway: measure first, then decide, because the switch is easy to flip but you give up every AI impression the moment it takes effect.