Google’s AI Search Opt-Out Feature: Understanding Its Implications for Publishers

| Editorial Team On 4 Aug, 2026

Google's AI Search Opt-Out Is Live — But the Cost of Using It Remains Unknown

Google is rolling out a Search Console setting that lets website owners remove their content from AI Overviews, AI Mode, and Discover's AI features without affecting regular search results. The catch: new tracking data suggests opting out may also pull sites from Top Stories placements appearing inside those AI surfaces.

The stakes are high enough that publishers and marketers need to understand exactly what the control covers before touching it. Born from regulatory pressure in the United Kingdom and a Google product initiative running in parallel, this setting represents the first time website owners have been able to separate AI feature inclusion from broader Search participation — but the data needed to make a confident decision simply does not exist yet.

What the New Search Console Control Actually Does

The setting lives under Settings in Google Search Console and is currently rolling out to a subset of website owners. Not every account has access yet.

The default option keeps a site included, meaning content can appear as links and help ground AI responses across AI Overviews, AI Mode, and Discover's generative AI features. Choosing to exclude removes the site from all of those features entirely — links included. Google said it began respecting these changes on June 17, with content typically dropping out within one to two days after processing, though caching can delay results.

Critically, the setting does not affect a site's performance in regular search results and carries no ranking signal. It also does not override separate participation choices in Merchant Center or Google Ads. Shopping participation remains an independent decision.

AI training is governed by a separate control entirely — the robots.txt token Google-Extended — which is a distinct lever that does not determine what content shows up in Search. Understanding the boundaries between these controls matters enormously before making any changes. For a broader foundation on how these systems connect, it helps to first understand how artificial intelligence works and why it's reshaping search.

Jamie Indigo, Director of Technical SEO at Cox Automotive, spotted the setting appearing on a U.S. account in July and flagged it on LinkedIn: "Search generative AI controls in Google Search Console. I'm not even British and it's not even my birthday!"

The Three Controls Publishers Are Confusing

One of the most consequential risks right now is treating Google's overlapping controls as interchangeable. They are not. Here is how they divide:

Google-Extended (robots.txt token): Governs whether Google's Gemini models can train on your content. Has no bearing on Search visibility.
Search Console AI toggle: Governs whether your content appears in AI Overviews, AI Mode, and Discover's generative features — including as links within those surfaces.
Robots.txt: Governs crawling only. Does not control indexing or AI feature inclusion independently.

Conflating these three will produce unintended consequences. A publisher who blocks Google-Extended believing it removes them from AI Overviews will be mistaken — and potentially invisble to Gemini without achieving the Search outcome they intended.

Why a UK Regulator Helped Shape This Feature

The Gap That Existed Before

Until this year, removing content from Google's AI features meant removing it from Search altogether. Tools like noindex and nosnippet existed long before AI Overviews but were never designed to separate generative features from traditional ones. Nosnippet, for example, removes content from AI Overviews but strips traditional snippets at the same time — an all-or-nothing outcome that left publishers with no middle ground.

Google acknowledged this gap in January, stating it was exploring ways to offer opt-outs specific to AI features. That statement landed the same day the UK's Competition and Markets Authority opened a consultation on requiring such controls.

In June, the CMA imposed a conduct requirement directing Google to give websites more control over how their content is used in generative AI. Google began testing the toggle with UK properties the same week. In the UK, compliance is mandatory with deadlines extending into 2027. The control began appearing on accounts outside the UK in July.

What Researchers Are Arguing

University of Oxford researcher Spencer Cohen and UCL competition-law PhD candidate Todd Davies argued in a paper published in the Journal of European Competition Law and Practice that this remedy may not go far enough. Davies, who the paper discloses worked at Google as a software engineer until 2022, wrote on LinkedIn summarising their position: "We argue that forcing Google to let websites opt-out of appearing in AI Overviews would be ineffective." Their case is that an opt-out does not protect publisher business models or create meaningful choice over how content is used.

This debate sits within a much larger shift in how search engines are being redesigned around AI. Publishers navigating these changes benefit from understanding search experience optimisation as a discipline — because the rules governing visibility are no longer purely about rankings.

The Top Stories Complication Publishers Need to Know

Where Google's "Separate Systems" Framing Breaks Down

Google's setting treats AI features and regular results as separate systems. The problem is that Google's actual results page does not always keep them separate.

NewzDash, which sells news-visibility tracking to publishers, released data showing Top Stories carousels rendering inside AI Overviews on U.S. trending news results. John Shehata, CEO and founder of NewzDash, put a number on the pattern: "Nearly 1 in 6 U.S. trending news queries now place Top Stories inside AI Overviews." The 15.5% rate applies to tracked results where Google displayed Top Stories. The UK figure is 17.46%. NewzDash has not published sample sizes or collection dates alongside the figures.

Kyle Sutton, Head of SEO and AI Discovery at The Washington Post, noted the same behaviour anecdotally: "Anecdotally, seems we're all seeing it a lot more often."

Shehata connected the pattern directly to the new control: "Using Google's newer Search Console generative AI opt-out is different, and will likely remove publishers from Top Stories inside AI Overviews." Google has not officially confirmed that interpretation. Google's help page states that excluded sites won't appear in AI features but does not address how the control handles a traditional feature rendered within an AI surface.

This is the central unresolved question for news publishers specifically. If Top Stories carousels are being served inside AI Overviews at scale — and the opt-out removes sites from AI Overviews entirely — then the practical effect of opting out extends beyond what Google's documentation describes.

The Business Case Against Opting Out

Rahul Jain, CEO and co-founder of Noble, argued against opting out earlier this year: "Opting out of Google's AI Overviews will hurt most publishers more than it helps." His reasoning is that exclusion removes websites from where user attention is concentrated. Noble sells services designed to help brands appear in AI-generated answers.

The generative AI performance report — also rolling out to a subset of accounts — shows impressions from AI features by page, country, device, and date. It does not include clicks or query data. Vahe Arabian, founder and editor-in-chief of State of Digital Publishing, described the practical workaround: "The job isn't picking a favourite dashboard; it's blending them into one scorecard."

The CMA requires Google to provide click-through data and additional publisher tools by December. Page-level controls for generative Search features are required by March 2027. Google must also report on compliance every six months during the first year.

What to Do Before Making Any Decision

NewzDash has said it plans to test the opt-out effect directly. Until that data is available, the decision carries meaningful risk in both directions — opting in leaves publishers exposed to potential content use without measurable return, while opting out may remove them from surfaces where audience attention is actively concentrated.

For publishers and marketing teams managing search visibility across both traditional and AI-driven surfaces, familiarity with the full suite of Google tools available to website owners is increasingly essential — the landscape of controls is expanding faster than most teams have adapted to it.

Three ways to use this information before making any decision:

Audit your AI feature exposure first. Check the generative AI performance report in Search Console to see which pages are earning impressions in AI Overviews and AI Mode before considering any change.
Map which control governs what. Google-Extended governs Gemini model training. The Search Console toggle governs AI feature inclusion in Search and Discover. Robots.txt governs crawling only. Treating them as interchangeable will lead to unintended consequences.
Wait for click data before committing. The CMA has mandated that Google deliver click-through metrics by December — that data will provide the first real baseline for calculating whether AI feature visibility is driving meaningful traffic to your site.

For further context on Google's evolving approach to AI in Search, the UK Competition and Markets Authority's published findings on Google's AI conduct requirements offer the most authoritative public record of what is being mandated and on what timeline.