Publishers and Google account holders can now block their content from feeding Google’s generative search features, after regulators forced Google to add new controls and Google began rolling account-level privacy settings. The Competition and Markets Authority required Google to offer effective opt-outs and clearer attribution, and Google has started testing a Search Console toggle with a subset of U.K. publishers while adding a Search Services History option to individual My Activity pages. For publishers the change means sites can be removed from AI Overviews and other generative features without losing standard search ranking, and for signed-in users it means uploaded images, audio and video can be prevented from being saved for model training. The CMA gave Google a compliance timetable.

Sites can be removed from AI-generated summaries because the Competition and Markets Authority required Google to provide publisher opt-outs and clearer links back to source content.

1. What the new controls do, in plain terms

Publisher opt-out lets a verified site owner stop pages from appearing inside AI features such as AI Overviews, AI Mode, and AI Overviews in Discover. Google describes the setting as a toggle in Search Console, the free webmaster toolset that verifies site ownership and manages how properties appear in Search. During initial testing Google rolled the toggle to a subset of U.K. publishers and says it intends to expand availability globally after testing.

Google has said sites that opt out won't receive impressions or traffic from generative AI features, and that opting out won't change how those pages rank in ordinary search results. The CMA expressly forbade Google from using a publisher’s opt-out decision as a ranking signal for standard search.

Account-level controls appear inside a new Search Services History tab on the My Activity page for signed-in users. The control aggregates what Search saves, and covers a wider set of data than text queries. It includes images uploaded for Lens searches, files uploaded to Search, voice searches and audio or video captured during interactions such as Search Live or Translate practice. Users can switch the setting off to stop future uploads being saved for model training, and they can delete saved items from their account view.

2. Publisher procedure: five steps to apply the opt-out

First, verify control of your site in Search Console. Only an account that has completed the verification flow can apply publisher-level settings.

If a newsroom, blog or commercial site lacks Search Console access, assign an owner or an authorised user and finish the verification before you try the opt-out.

Second, find the AI controls for your property once they appear. Google says the toggle will be visible inside Search Console for properties included in the test, and documentation will indicate whether the control applies at the property level or with other scoping options.

Third, understand the immediate effects. When a site is opted out, Google states the site won't receive impressions from generative AI features and its content won't be used for AI fine-tuning under the CMA-mandated rules. That means, for example, an article that previously appeared as a source in an AI Overview should stop being used in AI responses after the opt-out takes effect. The CMA also made clear that the opt-out can't be used to downrank sites in standard search results.

Fourth, review the new AI metrics in Search Console before you decide. Google has started to surface metrics that report which pages appear in AI responses and in which countries those appearances occur. Publishers should use those figures to measure current impressions and engagement from AI features so they can judge the traffic trade-off before switching the toggle.

Fifth, track compliance deadlines and document any communications. The CMA set a compliance timetable and expects Google to publish compliance reports that include supporting metrics.

Mastheads

If your outlet is negotiating content licensing, note that the regulator framed the opt-out as strengthening publishers' negotiating position. Keep records of implementation timelines and any Google messages visible in Search Console.

Worked example for publishers: a local news site has limited Search Console access and sees its long-form features appearing in AI Overviews for several countries. The editor first assigns a verified owner, checks the AI metrics panel to quantify impressions from AI features, then flips the toggle for the property. After switching off the setting the team continues to monitor Search Console metrics and keeps a log of dates and screenshots to support any future discussions with Google.

3. Account-level steps for individual users

Signed-in users get a separate control and a different workflow. Start by signing into the Google account you use for Search and go to My Activity, where a new Search Services History tab aggregates what Search retains.

Review what's saved. The My Activity view shows queries, uploaded media and any associated recordings, so you can identify specific images, audio or files you want to remove. Early reporting from account tests shows the media-saving option was on by default for some accounts, so it's worth checking proactively.

Turn off the Search Services History toggle and uncheck any Save media box to prevent future uploads from being retained for model training. Google’s public comment framed these settings as optional controls users can toggle at any time and as helping to get more relevant results.

Delete saved activity if you want it removed from your account view. After switching the setting off, use the delete functions in My Activity to remove prior entries so they no longer appear in your account history. That action clears the item from your account, though separate retention rules may still apply to material that has already been disconnected and used in training.

Factor in training-data retention. An in-product message observed in early testing indicated that training data might be kept for a period of time after it's separated from the user account. That creates a window during which media you uploaded could remain in Google’s training corpus even after you turn settings off or delete activity from My Activity.

Worked example for users: a photographer who used Lens to identify plants finds several uploaded images in My Activity. They switch off Search Services History, uncheck the Save media option, then delete the previous uploads from their activity log. The photographer understands this stops future use immediately, but notes the in-product retention message and, if concerned, follows Google’s published documentation or contacts support for further clarification.

4. What the controls don't do, and the limits to be aware of

Opting out has clear limits. For publishers, the opt-out removes content from generative AI features and prevents that content being used for fine-tuning under the CMA rules, but it doesn't change how a page ranks in ordinary Search. For individuals, disabling Search Services History and unchecking Save media prevents future uploads from being retained for model training and lets users delete previously saved activity from their account view.

However, training-data retention creates an important caveat. Google indicates that when saved media is used to train its models the training data is disconnected from the account and may be retained for a period after deletion. That means removing an item from My Activity may stop further use but may not erase material already incorporated into a model. Early reporting recorded an in-product message that training data could be kept for a period of time after it's separated from the account.

There is also divergence in public reporting about rollout timing and scope. All sources agree Google will provide publisher opt-outs through Search Console and that Search Services History is available for account-level control, and all agree the CMA set a compliance timetable. But accounts differ on exactly when controls will be available outside the U.K. test group and on how soon publishers can expect full functionality. The most concrete regulatory milestone remains the CMA compliance deadline and its requirement that Google publish reports documenting compliance with supporting metrics.

5. Practical tips and a decision checklist

First, map who in your organisation has Search Console verification and access. If you lack a verified owner, assign one now. Second, use the AI metrics Google has added to Search Console to quantify present exposure in AI features before flipping any opt-out. Third, if you opt out, plan for the loss of impressions from generative features and build alternate traffic plans.

Publishers negotiating commercial licences should record dates and messages visible in Search Console and lean on the CMA’s framing of the opt-out as strengthening publishers’ bargaining position. Individuals who need certainty about long-term removal should follow Google’s account deletion flows and consult Google documentation or support for clarification about retained training data.

Worked decision scenario: a specialist publisher that earns referral revenue from AI features weighs keeping their content available to generative responses. They check the Search Console AI metrics to see which articles drive most AI impressions, estimate expected traffic loss if they opt out, and decide whether the licensing power gained by opting out outweighs immediate referral revenue.

In short: if you manage a site, verify Search Console and read the new AI metrics before you act. If you are a signed-in user, check My Activity, turn off Search Services History and delete saved media if that matters to you. Keep in mind that items already used for training may remain in models for a period of time after separation from your account.

Related Articles

The CMA has set a compliance timetable and expects Google to publish reports with supporting metrics. Those reports, and the wider Search Console rollout, are the next milestones to watch: they will show when publisher opt-outs reach beyond the U.K. test group and whether Google’s retention practices meet the regulator’s requirements.

This article was created with AI assistance.