Brussels has asked publishers whether Google’s AI-search opt-out offers a meaningful choice. That turns an apparently technical control into a live question of traffic governance: can a publisher decline AI use of its work without giving up the visibility needed to sustain the work in the first place?
For people doing GEO work, the important shift is not merely that regulators are looking. It is that the trade-off between appearing in AI-generated answers and preserving publisher traffic is now being examined as a possible competition issue. Google introduced a global opt-out for its AI search features in June, while maintaining that ordinary search rankings would not be affected, according to thenextweb.com. But ordinary rankings are not the whole question when AI-generated answers occupy the part of the results page users encounter first.
The regulator’s question is narrower—and more consequential—than it looks
The European Commission sent publishers a questionnaire in July, with responses due on August 28. It asked whether they intended to use Google’s opt-out, what would cause them to use it, and how they viewed Google’s May decision to combine AI Overviews and AI Mode into a single AI Search experience, as reported by thenextweb.com.
Those are practical questions, but they go to the core of whether an opt-out is a remedy or simply a mechanism for assigning risk to publishers. A control can exist in product terms and still be economically unusable. If declining AI inclusion means becoming less visible in the search surface where attention is concentrated, then publishers may be choosing between unattractive outcomes rather than exercising meaningful control.
The available reporting describes that choice starkly. Remaining in AI search can mean allowing reporting and analysis to be used in summaries that satisfy a reader’s query without a click. Opting out can protect content from those summaries, but may reduce visibility in an increasingly prominent part of search, according to europesays.com. The regulatory question is therefore not whether a setting exists. It is whether publishers can realistically use it.
Why this matters to GEO work
GEO is often discussed as the work of making content legible, citable and useful inside answer systems. For publishers, however, greater answer-system usefulness can sharpen the commercial tension. The more completely an AI response resolves a reader’s need, the less certain it is that the reader will visit the source page.
That is not an argument against making content clear or structured. It is a reason to distinguish visibility from value capture. A brand, article or source may be present in an AI experience while receiving little measurable referral value from it. The supplied reporting says publishers have reported click declines of up to 58%, though it does not establish a universal rate, a common measurement method, or a causal figure for every publisher or query type. That limit matters: teams should not treat 58% as a forecast. It is evidence that the traffic concern is substantial enough to have entered antitrust litigation, as thenextweb.com reports.
For GEO practitioners, the Commission’s questionnaire points to a more rigorous operating question: what evidence would demonstrate that AI inclusion is beneficial, neutral or harmful for a particular publisher? “We appear in AI answers” is not enough. Neither is “we have retained organic rankings.” The relevant comparison is between the value of AI-search exposure and the referral, subscription, advertising or other publisher value that may be displaced when an answer resolves the query before the click.
The merged AI Search experience makes separation harder
Google’s May decision to bring AI Overviews and AI Mode into a single AI Search experience changes the scope of the choice publishers are being asked to make. The reported concern is that merging the features leaves publishers with fewer distinct search-page components they can choose to join or avoid, according to thenextweb.com.
This is the critical design issue. An opt-out has more practical value when it can be applied to a distinct use while preserving participation in other useful forms of discovery. It has less value when the consequence is effectively participation in both AI summaries and search visibility, or withdrawal from both. remoteworkeurope.eu characterises the current position as publishers being able to choose only “both or neither.”
The sources do not provide the technical detail needed to say exactly how Google implements the opt-out, which content types it covers, or how its effects vary across queries. That evidence gap should constrain any operational conclusion. What is clear from the reporting is the issue Brussels is investigating: publishers are being asked whether the control actually separates the AI-summary decision from the search-indexing and visibility decision.
What a publisher should be able to show
The Commission’s request gives publishers a reason to turn general dissatisfaction into a documented account of choice and consequence. A useful internal record would separate observed facts from inferred effects.
- Track the pages, topics and query classes where AI-generated answers are visible and where the publisher’s work appears to be used or represented.
- Establish organic referral baselines, then record changes alongside the growth or placement of AI-search experiences where those changes can be observed.
- Identify which content has the highest risk of being fully answered in a summary, rather than requiring a visit for depth, evidence or utility.
- Model the cost of each available choice: continued AI inclusion, reduced AI inclusion, and any resulting change in visibility or referral traffic.
- Keep a record of why an opt-out would be used, avoided or considered commercially impossible.
None of this proves that an AI feature caused a particular traffic change. The supplied sources do not provide a methodology that would support such a causal conclusion. But it does make the actual trade-off auditable. That is especially important when the regulator is asking not simply whether a tool exists, but what conditions would make publishers use it.
Compensation and control are different remedies
The reporting also makes clear that some publishers are seeking more than withdrawal rights. French media organisations have asked their competition authority to require Google to pay for the use of content in AI-generated answers, while a German court found Google liable over AI Overviews, according to thenextweb.com. The UK, Germany and France are described as examining related problems through different mechanisms: conduct requirements, courts and competition-authority action.
That matters because an opt-out does not answer the compensation question. A publisher may conclude that appearing in AI search remains necessary for reach while also believing that use of its content in answer generation should carry payment. Conversely, compensation would not automatically restore the ability to control how content is presented or whether a summary substitutes for a visit.
The underlying imbalance described by the reporting is structural: Google controls the search interface, ranking system and AI features, and therefore controls both the options offered and the visibility consequences attached to them. europesays.com reports that this helps explain why publishers are pressing for compensation rather than relying only on an ability to withdraw.
What happens next is not fully settled in the supplied reporting
There is a material inconsistency in the sources about the Commission’s procedural position. thenextweb.com says the questionnaire is not an investigation and may help Brussels decide whether there is enough evidence to open a formal competition inquiry. remoteworkeurope.eu, however, says the questionnaire tests whether there is enough evidence to escalate beyond a formal probe opened in December 2025. On these extracts alone, the formal status cannot be resolved.
That uncertainty does not weaken the practical lesson. The Commission has asked for evidence about whether Google’s opt-out is meaningful. GEO teams at publishers should treat this as a governance problem, not a visibility tactic alone. The decisive question is whether AI-search participation can be chosen on terms that preserve a viable route from discovery to the publisher’s own property. If the answer is no, the opt-out is not much of a choice.