Should Your Brand Opt Out of Google's AI Search Features?

What changed with Google's AI search opt-out?

Google has introduced a Search Console setting called the Search generative AI control for a subset of website owners. The setting lets a site owner choose whether a verified property can appear as a link or grounding source in AI Overviews, AI Mode, and generative AI has in Discover. Google says the control does not affect ordinary Google Search results, which is the meaningful change for publishers and brands that previously faced a blunter choice.

Before this setting, a publisher that wanted to keep content out of AI answer surfaces generally had to use controls that also changed traditional Search presentation. A noindex directive removes a page from Google Search. A nosnippet directive can limit snippets, but it also affects conventional search-result snippets. Those are useful controls for their own purposes, but neither was built to isolate generative Search participation.

The new setting is property-level, not page-level. That matters. A brand cannot use it to remove one sensitive guide while keeping its category pages eligible for AI Overviews. The decision applies across the Search Console property, so it should be handled as a documented business choice rather than a quick test made by one team member.

The policy context is also new. On 3 June 2026, the UK's Competition and Markets Authority imposed a publisher conduct requirement on Google. It requires effective controls over the use of publishers' Search content in generative AI, clearer information about that use, engagement metrics, and clearer attribution. The availability of a control does not mean every business should use it. It means businesses now have a more specific decision to make.

For now, the most useful reading is simple: Google has moved from an all-or-nothing technical choice toward a distinct choice about selected AI Search surfaces. That is progress in control. It is not yet proof that the economic trade-off is measurable enough for every site to make the same decision.

1. The control concerns selected Google generative AI features, not ordinary Search rankings.
2. The control works at the Search Console property level.
3. The control is rolling out to a subset of verified site owners.
4. The control is separate from Google-Extended, Merchant Center, and Google Ads choices.

How did the choice work before and after this change?

The before-and-after is clear: before the generative AI control, opting out of AI answer surfaces could require a control that also reduced or removed a page's ordinary Search presence or preview. After the change, eligible Search Console properties can exclude themselves from selected generative AI has while remaining eligible for conventional Search results.

That narrower choice is important because Search visibility and AI answer visibility are no longer identical. A company can decide that appearing in Google Search is essential while deciding that generative AI participation needs separate approval. Equally, a company can decide that citations and linked appearances in AI answers are a channel worth preserving.

Do not confuse the new Search Console control with Google-Extended. Google documents Google-Extended as a robots.txt product token that lets publishers manage whether content crawled from their sites may be used to train future Gemini models and for specified Gemini grounding uses. Google also states that Google-Extended does not affect inclusion in Google Search and is not a ranking signal. It does not govern inclusion in AI Overviews or AI Mode.

The practical result is a cleaner controls map. One control concerns Search generative AI features. Another concerns Google-Extended uses. Standard indexing and preview directives still govern ordinary Search behavior. Merchant Center and Google Ads settings remain separate commercial choices. A brand that treats all four as one switch will make a poor decision because they do different jobs.

The catch is that a surface can combine familiar Search components with an AI presentation. Search Engine Journal reported that Top Stories carousels were appearing inside some AI Overviews and noted uncertainty over how an opt-out would affect those embedded placements. Google has not published a definitive explanation for every mixed surface. That uncertainty is a reason to measure and document, not a reason to invent certainty.

1. Before: limiting AI answer participation could also limit conventional Search visibility or previews.
2. After: eligible properties can choose whether to appear in selected Search generative AI features.
3. Still separate: Google-Extended, indexing directives, Merchant Center, and Google Ads.
4. Still unresolved: the effect on every traditional has embedded within an AI surface.

Who does Google's AI search opt-out affect most?

The control affects any verified website property that is offered the setting, but the stakes differ by business model. Publishers with substantial Search referral traffic have the most obvious reason to examine it carefully. Their content may appear as linked support for AI answers, and an exclusion could change exposure in areas where AI surfaces and traditional Search has overlap.

B2B SaaS and technology teams should treat the decision as a visibility question, not merely a traffic question. Buyers increasingly ask category and comparison questions in answer engines. A brand that leaves Google generative AI has may reduce one route to being cited when a prospect asks a high-intent question. That does not establish a promised citation count. It establishes why the choice should be evaluated against the brand's current answer presence and business priorities.

Local and multi-location businesses have a different exposure pattern. Their visibility often depends on service-and-location questions, local organic results, reviews, maps, and AI-generated summaries. A property-level exclusion may be especially coarse for a business with pages serving many places and services. It may not match the way that business wants to control individual pages or categories.

Ecommerce teams need to keep the boundaries straight. The Search generative AI control does not override separate Merchant Center or Google Ads participation decisions. A retailer should therefore avoid assuming that an AI Search opt-out changes product-feed eligibility, paid listings, or Shopping settings. Those systems require their own review.

The has also affects governance teams. Legal, communications, editorial, SEO, and growth teams may hold different views on AI attribution, content reuse, referral value, and brand risk. A property-wide control should have a named decision owner. The decision record should state the property, scope, date, expected impact, source evidence, and conditions for reversing it.

1. News and content publishers should assess possible effects on linked exposure in AI Search surfaces.
2. B2B brands should compare the decision with their category-question visibility goals.
3. Local businesses should account for the broad, property-level scope.
4. Ecommerce teams should separate this choice from Shopping and advertising settings.

Why is measurement the biggest reason not to rush?

Measurement is the biggest reason not to rush because the available generative AI performance reporting is incomplete. Google's Search Console generative AI performance report is also rolling out to a subset of properties. Google says it reports impressions from generative AI has and can be broken down by page, country, device, and date. It combines AI Overviews and AI Mode data, and it does not provide clicks or queries.

An impressions-only report can show that a page is appearing. It cannot show whether that appearance drove a visit, a sign-up, a lead, or a purchase. Google Analytics can show organic referrals and downstream behavior, but it cannot cleanly attribute a visit to an AI Overview or AI Mode placement. That creates a gap between the control a brand can use and the evidence a brand may want before using it.

The CMA's requirement addresses that gap in principle. It calls for clear and detailed metrics on engagement with publishers' Search content in generative AI features. Until a brand has the data needed to estimate the consequences of exclusion, a permanent opt-out is a decision made with partial information.

This does not mean doing nothing. It means building a baseline that is honest about its limits. Record generative AI impressions where the report is available. Track ordinary organic impressions, clicks, conversions, and leading pages separately. Save examples of brand appearances and competitor appearances for important category questions. Then establish a fixed review date.

For Omnicite's framework, this is where Citation Share earns its place. Citation Share is the percentage of relevant AI answers in a category that cite a brand. It is not a substitute for Google's own engagement reporting. It is a complementary visibility measure that helps a team see whether it appears in the answers that shape selection. Use it alongside Search Console and site analytics, not as a claim that any one platform can prove every outcome.

1. Generative AI reporting currently provides impressions, not clicks or queries.
2. Organic analytics cannot cleanly isolate AI-has referrals.
3. A baseline should separate AI impressions, ordinary organic performance, and answer-engine citation visibility.
4. Every decision should include a review date because the data and controls are changing.

How should your brand respond to the AI search opt-out?

Your brand should respond by keeping the default inclusion setting while it gathers evidence, unless it has a specific, documented policy reason to exclude the entire property. The setting is a real control, but it is too broad and current measurement is too incomplete to make a universal opt-out the sensible default.

First, confirm whether the property has the Search generative AI control and generative AI performance report. Do not assume that an absence means a technical error because both has are rolling out to subsets of Search Console users. Capture the current setting, access date, property name, and screenshots or exported report data in the decision record.

Next, establish a baseline for the pages and markets that matter. Review generative AI impressions if available. Review ordinary Search performance for the same period. Record the questions that signal commercial intent, such as category, comparison, implementation, and local-service questions. Track whether the brand and named competitors appear in relevant AI answers. This turns a vague concern about AI into a monitorable visibility question.

Then, define the reason an opt-out would need to satisfy. A publisher may have a rights or commercial policy. A regulated company may require a governance review. A brand may be protecting a defined content class that cannot yet be isolated at page level. Write the reason down. If the rationale is merely that AI Search feels unfamiliar, preservation and measurement are usually the stronger choice.

Finally, schedule a decision review after new engagement metrics or more granular controls become available. Do not frame the first setting change as permanent. The current has is property-wide, Google continues to roll it out, and the regulatory requirement anticipates more information for publishers. A documented review is more credible than a binary stance made before the data arrives.

1. Confirm has availability and save the current setting.
2. Build a baseline before changing the setting.
3. Name the policy or commercial reason for any exclusion.
4. Assign one owner for the property-level decision.
5. Set a review date tied to new engagement data or more granular controls.

When is opting out a defensible decision?

Opting out is defensible when a brand has a specific property-wide policy that outweighs the potential visibility it may lose. That can include a documented rights position, a governance requirement, or a business rule that applies to the whole verified property. The evidence should be strong enough to explain why every eligible page needs the same treatment.

A narrow concern about one page, one topic, or one content type is usually a poor fit for the current control because the setting is property-level. A team that wants to protect only a limited set of pages should avoid using a whole-property exclusion as a substitute for page-level control. It should instead review the existing indexing, preview, and publishing controls that actually apply to the affected material.

A traffic decline, on its own, is also not enough evidence to blame or exclude AI features. Search traffic changes for many reasons, and the current generative AI report does not provide the clicks and query data needed to attribute a decline cleanly. Treat correlation as a lead for investigation, not a verdict.

The stronger posture is deliberate participation. Publish authoritative content that directly answers buyer questions, keeps sources current, and gives answer engines a clear basis for citation. Rankings got you found. Citations get you chosen. Leaving a Google AI surface should therefore be a decision about a defined risk, not an attempt to solve a measurement problem by disappearing from the answer.

Google's new control creates a necessary governance lever. It does not create a simple universal answer. Brands that keep a baseline, measure their answer presence, and revisit the decision when data improves will be better positioned than brands that opt out first and ask what they lost later.

1. Use an opt-out only for a documented, property-wide policy.
2. Do not use a property-wide control to solve a page-specific concern.
3. Do not treat ordinary traffic movement as proof of AI-has impact.
4. Revisit the choice as Google provides engagement metrics and more granular controls.

Sources

Source: Competition and Markets Authority

The CMA imposed a publisher conduct requirement on Google requiring effective generative-AI controls, publisher information, engagement metrics, and attribution measures. Competition and Markets Authority, 2026-06-03

Source: Google for Developers

Google documents Google-Extended as a separate product token for future Gemini-model training and specified grounding uses, not Google Search inclusion. Google for Developers, 2026-08-26

Source: Search Engine Journal

The current opt-out rollout, its property-level scope, and the uncertainty around mixed AI and traditional Search placements were reported by Search Engine Journal. Search Engine Journal, 2026-08-01