The toggle that removes your content from AI Overviews and AI Mode is here. Since 31 August 2026 every property worldwide has it, including yours. According to Google, opting out costs you nothing in normal rankings. What used to be a waiting game is now a decision.
- Google enabled the toggle for all websites worldwide on 31 August 2026. Before that it ran as a test, first for a subset of sites in the United Kingdom, from July also outside it. The report that shows where you appear in AI features went global on the same day.
- In Search Console the setting is called “Search generative AI”. Without its own setting, a property takes the value of the nearest parent, at the top the domain property. Child properties take over its value as long as the box “Inherit control from” is ticked.
- For most sites the opt-out is not a good idea. A serious case exists mainly for subscription and premium publishers as well as brands with a concrete reputation risk.
- Do not confuse the four levers: the new toggle,
Google-Extended,nosnippetandnoindexeach do a different job. - In May 2026 the Regional Court of Munich I classified Google as a direct interferer for false statements in AI Overviews. The ruling never became final: after Google appealed, the parties settled. The liability question therefore remains open.
Since AI Overviews launched, the deal was one-sided. Your content ended up in the AI answer, whether you wanted it or not. The only ways out were snippet blocks such as nosnippet, which leave your normal search result without a text snippet, or noindex, which takes you out of search entirely. For a site that lives on search traffic, that was no real choice.
That is over. Google enabled the opt-out toggle in Search Console for all websites worldwide on 31 August 2026. It removes your content from the generative AI features of search and is, according to the Search Console help, explicitly not a ranking signal for the rest of search. Sounds like what publishers have been waiting for since AI Overviews launched. It is. With two big asterisks.
First, it bundles three functions that you would rather control separately. AI Overviews, AI Mode and the generative features in Discover all hang on a single switch. Second, you still decide without click data, because the new report shows impressions as its only metric, no clicks. Then there is the German special case. In its ruling of 28 May 2026 the Regional Court of Munich I classified Google as a direct interferer for false statements in AI Overviews. After Google appealed, the case ended in a settlement, and the ruling never became final. The regulatory and the legal front are moving at the same time anyway. Both change the question of whether you want to stay inside AI Overviews.
Important for context, because two different things get mixed up here: this article is not about browser extensions that let end users hide the AI overview in their own search. It is about operator control. Your website should no longer be used and linked as a source for the AI answers. That is an SEO decision, not a browser setting.
Since the rollout the toggle sits in every property, including my own. That is exactly where the problem starts. Out of the box it is set to include and cascades down to the child properties. Nobody chose that default, Google set it. That is why the setting now belongs in a technical audit. On top of that comes the old risk of confusion: four levers sound similar, three of them do something different from the new toggle. That causes mistakes as soon as a wrong setting cascades across an entire property.
What the “Search generative AI” control really does
Google officially calls it the Search generative AI control. In Search Console you find it under Settings → Search generative AI. In the German interface the same item is called Einstellungen → Generative KI-Funktionen der Google Suche. According to the official Search Console help it manages your site’s visibility in three areas: AI Overviews, AI Mode and, according to the documentation, also generative AI features in Google Discover. Google itself notes that it expects to update this list over time. What is switched on and off here I broke down in detail in the pillar on the difference between AI Mode and AI Overviews. This post is the spoke to it: there the what, here the whether and how of the exclusion.
The settings page shows two options. For a child property a checkbox comes first:
| Option in Search Console | Effect |
|---|---|
| Include my site’s links and content in Search generative AI features (Default) | Your content can appear in the AI features, be linked and contribute to the AI answer. You can receive impressions and traffic from these features. Default for all properties. |
| Exclude my site’s links and content from Search generative AI features | Your content no longer appears in the AI features. It is not linked there and not used for generation. No traffic and no impressions from these features. |
| Inherit control from: <parent property> (checkbox) | The property follows the setting of its parent property. That is the default as soon as a parent property exists. If you untick the box, you choose yourself. |
The decisive sentence is in Google’s documentation in black and white: this control is not used as a ranking or inclusion signal for other parts of search. According to Google, an exclusion therefore costs you no position in the classic blue links. Your content may still feed into search’s general language understanding. Other Google services such as Merchant Center or Google Ads remain unaffected.
One technical detail matters in practice: according to the Google help, a change generally takes a few days: content is excluded within one to two days after the control goes live. Due to caching and propagation across Google systems it can take longer for some content. Google does not say how much longer. The current documentation names no restriction for a test phase.
Worldwide since 31 August: what changes for you
Google rolled the toggle out first only to a subset of website operators in the United Kingdom, explicitly for thorough testing. From July it also appeared outside the United Kingdom, among others in France. Since 31 August 2026 the staggering is over. The Search Console help now contains the sentence many had been waiting for: “As of August 31, 2026, we’ve rolled out this control to all websites worldwide.” (retrieved on 11 November 2026) On the same day Google made the report on visibility in AI features available globally. The two belong together, because without the report you decide blind.
The background is the British competition authority CMA. Under the Digital Markets, Competition and Consumers Act 2024 it obligated Google to give publishers effective control over how their content is used in AI features shown to users in the United Kingdom. In its statement of 3 June 2026 it gives Google “nine months to implement all changes”. According to the CMA notice, the requirement comes into force six months after the notice, so from early December 2026, with one part after nine months. Google does not explicitly present the toggle as its implementation. The CMA did not require Google to enable it worldwide.
For you this means one very concrete thing. As long as nobody has changed it, the control is set to Include. In child properties it is set to “Inherit control from”, and Search Console shows the inherited value there under “Current control”. Include is the default for all properties. So you did not actively decide to appear in the AI answers, it simply continues. What is new is that you could change it now. Another owner could change it too. The owner of a child property can also override it for their part.
That the data basis in GSC suddenly becomes a lever is not a new theme. Back when Google disabled the num=100 parameter, many had to learn to read their impression figures anew. With the AI report the same reflex applies: first understand the data basis, then interpret.
Opt-out ≠ Google-Extended ≠ noindex: separating four levers cleanly
Google-Extended, nosnippet and noindex do four different jobs. Confusing them either blocks too little (training keeps running) or too much (snippet gone, page out of search). Google separates this in its own documentation.
This is the point where many texts mix up the levers. Until this toggle there was no lever that only controlled the AI display. Google’s help explicitly separates the toggle from Google-Extended and noindex; nosnippet is covered by the robots meta tag documentation. Summarised as my own overview:
| Lever | Out of AIO & AI Mode? | Stops AI training? | Snippet stays? | Stays in search? |
|---|---|---|---|---|
| Search generative AI control (new) | Yes | No | Yes | Yes |
Google-Extended |
No | Yes | Yes | Yes |
nosnippet |
Yes | No | No | Yes |
noindex |
Yes | No | No | No |
Concretely this means, with Google’s own words as evidence: the new toggle does not affect AI training. For “to limit training of the models” Google explicitly points to Google-Extended. To take your content out of search completely, you still need noindex. Google-Extended, in turn, limits training but does not pull you out of the AI Overviews display. nosnippet was, until now, the usual emergency exit from the AI features that kept the page in search. The price: your normal snippet disappears too. This coupling is exactly what the new toggle dissolves. The Search Engine Journal also points to this double effect of nosnippet.
On 21 October 2026 Google extended its overview “Control what you share with Google Search”. It now lists the toggle as a site-wide control next to the robots meta tag, data-nosnippet and the X-Robots-Tag.
There is a fifth lever, one layer down at the CDN level, since September 2026: at Cloudflare the “Block” option for AI training also locks out Googlebot. “Disallow AI Training” lets it through for search. That happens on the server, not at Google.
Property inheritance: the stumbling block with parent and child
Google’s help describes this part in a section of its own. In practice it is error-prone. The logic in brief: by default a property inherits its control from the nearest parent property that has actively stopped inheriting. If none of the parent properties has set anything manually, the setting of the top-level domain property applies.
An example from Google’s documentation: the child property https://example.com/business/cats inherits, by default, the setting of its domain property example.com. That does not hold if a closer parent property such as https://example.com/business/ was set manually. Owners of a child property can at any time either follow the parent or set their own configuration for the property and its sub-pages.
Why this easily goes wrong in practice: many manage a domain property plus various URL-prefix properties in Search Console. They like to flip the toggle at the most convenient level. It is easy to overlook that the setting cascades downwards. Since the worldwide rollout this is no longer a theoretical risk. Before, only selected sites had the toggle, first in the United Kingdom. Now you check per property whether it inherits or has its own setting. Only then do you set anything to exclude.
This is what it looks like in Search Console. Both screenshots come from my own properties for seo-kreativ.de, taken on 11 November 2026. The numbers belong to the list below.
What to check before you save
- Untick the box first. For a child property, “Inherit control from” (1) is ticked by default. While the box is ticked, you cannot click either option. “Save” stays grey as well. That is not a lock. Only without the tick do you choose yourself between include (2) and exclude (3).
- “Current control” is an inherited value. Below the checkbox it says, for example, “Current control: Include”. That is the value of the parent property. It is not a decision of its own for this property.
- Decide at the right level. The top-level domain property has no checkbox. If you set it to exclude there, the notice in Search Console says it affects all pages in the property, including child properties. Exempt are properties with their own setting and everything that inherits from them.
- Open every property separately. Domain property, variants with and without www and subdirectories such as
/blog/each have their own settings page. For each one, check whether the box is ticked and which value arrives. - Wait after saving. According to Google’s help, an exclusion takes effect within one to two days, for some content later. Note your starting value in the AI report beforehand.
Should you opt out? A decision framework
The data behind the debate is real, but no licence for panic. An Ahrefs analysis sees the desktop CTR on position 1 around 58 percent lower when an AI Overview is present (data from December 2025), though as a correlation and not as a proven cause. Pew Research data (US, March 2025, around 900 adults) show an 8 percent click rate on a classic search result when an AI Overview is present, versus 15 percent without. In its submission to the CMA, the British Publishers Association cited a 19 percent drop in click-through to academic reference services, a figure from Oxford University Press. In a non-representative X poll by SEO industry observer Barry Schwartz (just over 350 participants), around 33 percent said they want to exclude their content as soon as Google allows it. The figures come from different surveys, each with its own methodology. They are not directly comparable, but they still work as a directional indicator.
The criticism of the toggle itself is still serious. You should know it before you praise the toggle. Paul Bannister of Raptive calls the opt-out, in essence, a light switch while the power plant keeps running. Real control looks different. Stuart Forrest (formerly of Bauer Media) puts the core problem in a nutshell: without AI-specific click data you cannot judge whether AI Overviews bring you traffic or only cut it off. Chris Dicker (Candr Media Group) criticises that Google bundles AI Overviews and AI Mode instead of making each feature individually controllable. He feared that Discover would be added too. Google’s help has listed the generative AI features in Discover since June 2026; Discover itself remains unaffected.
Against the gut feeling, a sober self-test helps. These are exactly the questions that belong before any opt-out decision. Since 31 August this is no longer a planning exercise but a setting you change in an afternoon. More than one or two clear yeses argue for opting out. Otherwise you stay in:
- Business model: Does your site live on paid content or subscriptions whose value drops as soon as the AI summarises them for free?
- Reputation risk: Is there a concrete risk that a faulty AI summary of your content harms you (see the Munich case below)?
- Traffic dependency: How large is the share of visibility you lose in the AI answer? Can you absorb it?
- Data basis: Do you have solid figures that AI Overviews cost you clicks for your topics rather than bring reach?
- Alternatives: Do you have distribution channels beyond Google search that can cushion a loss of reach?
The German special case: Munich I ruling, appeal and settlement
While the industry discussed the opt-out, a German court drew a possibly more far-reaching line, which no longer stands after the settlement. In its ruling of 28 May 2026, the Regional Court of Munich I (case no. 26 O 869/26) prohibited Google from spreading certain statements about two Munich publishers in its “AI overview”. The court treated them as untrue factual claims or inadmissible expressions of opinion. The ruling was issued in preliminary injunction proceedings. The AI had falsely linked the plaintiffs to subscription traps and dubious business practices. According to the court, some of these statements appeared in none of the linked sources.
What stood out legally was the classification: the court saw Google not as a merely indirect but as a direct interferer (“unmittelbarer Störer”). The AI overview summarises sources “in its own words and according to its own structure”. It thereby makes “its own statements, put forward by the defendant itself”. According to the court, the liability privilege of classic search engines does not apply here, not even via the Digital Services Act. Under that privilege an operator only makes third-party content findable. According to the ruling, Google was to pay 80 percent of the court costs and of the plaintiffs’ lawyers’ fees. In case of violation a penalty of up to 250,000 euros was threatened. The court explicitly refused to limit the injunction to Germany and referred to recognition under the Brussels Ia Regulation.
The ruling never became final. Google appealed. Before the Higher Regional Court of Munich (case no. 18 U 1744/26) the parties ended the dispute with a settlement, as noted in the official Bavarian database Bayern.Recht. The ruling is therefore moot. Bayern.Recht does not state when the settlement was reached or what it covers (as of 11 Nov 2026). According to the trade press, the Munich decision built on a ruling by the Regional Court of Frankfurt am Main from September 2025 (case no. 2-06 O 271/25). No decision by a higher court on liability for AI overviews was known as of 11 November 2026.
According to OnlineMarketing.de, SISTRIX boss Johannes Beus classified the ruling at the time as a possible directional decision. Google’s statement that it considers the decision not final is documented there as well. Christian Kunz of SEO Südwest rated it as significant. For practice this means: in Germany the law can be a second instrument when an AI overview harms you. A higher court has not settled this. How robust Google’s quality mechanisms are I covered in the post on how Google rates AI texts.
Infographic: the four levers compared
Frequently asked questions (FAQ)
Does opting out of AI Overviews cost me my rankings?
According to Google, no. The Search Console help explicitly states that this control is not used as a ranking or inclusion signal for other parts of search. Your classic organic positions and the Discover feed remain unaffected. Only the AI features then deliver no impressions and no traffic.
Can I use the opt-out toggle in Germany?
Yes. Google rolled the control out to all websites worldwide on 31 August 2026, including Germany, Austria and Switzerland. You find it in Search Console under Settings in the “Search generative AI” section. Before that it ran as a test, first for a subset of sites in the United Kingdom, from July also outside it.
Why can’t I select the options in Search Console?
Because the property takes over its setting from a parent property. For child properties the box “Inherit control from” is ticked by default. Then both options and the “Save” button are greyed out. If you untick the box, you choose yourself for this property. If you want to decide for all inheriting properties at once, change the domain property. It has no checkbox. Anyone who has unticked the box on a child property keeps their own choice.
Is the opt-out the same as Google-Extended?
No. Google-Extended controls AI training, not the display in search. Content can still appear in AI Overviews despite Google-Extended. The new toggle, conversely, controls the display and grounding in AI Overviews and AI Mode. It does not affect training. Google itself points to this separation in its documentation.
Which features does the control actually cover?
Google’s documentation names three: AI Overviews, AI Mode and the generative AI features in Google Discover. Google adds that it expects to update this list over time. Anything not on that list is not governed by this control. For training the models, Google explicitly points to Google-Extended.
Do I get enough data to make the opt-out decision?
Only partly. The report has also been available worldwide since 31 August 2026. It shows impressions, affected URLs, countries, devices and the trend over time. Clicks and queries are not in it. So you see where you appear in AI features, not whether that brings traffic. Exactly this gap is criticised by experts in the industry.
Does the opt-out also apply to Google Discover?
According to Google’s documentation, yes for the generative AI features within Discover, no for the normal Discover feed. You therefore disappear from the generative AI features in Discover but remain a regular Discover source. That matters, because many publishers use Discover as a standalone traffic source.
Conclusion: the toggle is here, so is the data gap
My advice has turned with the rollout. This used to be a preparation topic. Now it is a to-do, a small one. Check how the control is set in your properties. Start with the domain property and check every checkbox below it. Note the AI impressions of your most important URLs as a starting value. Keep the four levers apart. After that you decide in peace. For the vast majority of sites the decision is: stay in. The difference from before is that it is a decision.
The more interesting movement in Germany is coming from two sides anyway: the opt-out the British CMA required of Google for the United Kingdom, and the liability question from the Munich case. Both shake the same question: who does the AI answer belong to? This question will keep us busy for a long time.
Google-Extended, nosnippet and noindex cleanly apart. 4) Go through the decision framework and when in doubt, stay in.
Sources (retrieved on 11 Nov 2026)
- Google Search Console Help: Search generative AI control
- Google Search Central: Control what you share with Google Search
- Google Search Central: Latest Google Search Documentation Updates
- Google Search Central Blog: Report on generative AI features
- Google Crawling Infrastructure: Google-Extended
- Search Engine Journal: Google Gives Sites AI Search Opt-Out But Not The Data To Use It
- SEO Südwest: Gericht: Google haftet für falsche Informationen in AI Overviews (German)
- OnlineMarketing.de: Wichtiges Urteil: Google haftet für Falschaussagen in AI Overviews (German)
- Digiday: Google’s AI opt-out leaves publishers with a difficult choice
- Bayern.Recht: LG München I, judgment of 28 May 2026, 26 O 869/26 (German)
- CMA: CMA secures fairer deal for publishers and Publisher Conduct Requirement Notice
- Cloudflare: Accountable mixed-use AI crawlers
As of: 11 November 2026. The primary source is the official Google Search Console help on the “Search generative AI control”. It records the worldwide rollout as of 31 August 2026. The screenshots show the settings page of the seo-kreativ.de properties on 11 November 2026. The Munich ruling (Regional Court of Munich I, judgment of 28 May 2026, case no. 26 O 869/26) never became final. After Google appealed, the parties settled before the Higher Regional Court of Munich (case no. 18 U 1744/26), and the ruling is moot (research as of 11 November 2026). The statements on the CMA deadline are based on its announcement and notice of 3 June 2026. Statements on Search Console features are based on Google’s publications and reflect the documented status. They may change. All information without warranty and without claim to completeness. This article does not replace individual legal or SEO advice. All brand and product names mentioned are the property of their respective owners. Christian Ott, seo-kreativ.de


