Google confirmed on June 3 that it will comply with a UK Competition and Markets Authority requirement to let publishers opt out of having their content aggregated into AI search features. The mechanism is a toggle inside Google Search Console, the free tool publishers already use to manage indexing. Once opted out, a site will not appear in AI Overviews, AI Mode, or AI Overviews in Discover, but it will continue to rank in traditional Google search as before. The CMA called the remedy a world first.
The numbers behind the dispute explain why this matters. Google said AI Overviews now reaches more than 2.5 billion monthly active users and AI Mode crosses one billion. That is roughly the same order of magnitude as classic Google Search itself, and the share of queries answered without a click through has been rising steadily for two years. Publishers have been arguing that the aggregation is destroying their referral economics while training data is being scraped without compensation. The CMA, which designated Google with strategic market status in October 2025 and issued specific conduct requirements in January 2026, is now operationalizing that designation in a way that creates real bargaining power.
The decoupling from ranking is the part that gives the remedy teeth. In every previous standoff between Google and publishers, from Spain's link tax to Australia's media bargaining code, Google's implicit threat was that any publisher pushing back too hard would see its visibility decline. By forbidding any ranking penalty for opting out, the CMA has stripped that threat. A news organization can now refuse to feed AI Overviews while preserving its traditional traffic, which means the negotiation over content licensing happens on something closer to equal terms. Google is clearly aware of the dynamic, which is why it is rolling out a parallel set of metrics in Search Console showing AI impressions, which pages appear in AI responses, and country level data, to make it harder for publishers to walk away from the visibility on offer.
For CTOs at media businesses, the immediate action is straightforward, run a controlled experiment. Opt out a subset of titles, measure referral traffic, measure brand search lift from continued ranking, and put a number on the actual cost of AI aggregation versus the contribution it makes to discoverability. We have been having a version of this conversation with retail and grocery clients whose product content has been showing up in AI Overviews without driving conversions. The same toggle, once it goes global, will let them test the same hypothesis on product pages, reviews, and how to content.
For CTOs at non media businesses, the strategic implication is broader. The CMA has shown that a strategic market status regime can produce a precise, technically enforceable remedy on a short timeline, less than nine months from designation to working opt out. Brussels under the Digital Markets Act, Washington under the various antitrust suits, and Tokyo under its own ex ante framework will all study this template. Vendors and platforms that rely on harvesting other people's content or data to power AI features should assume that opt out mechanisms with similar properties, no retaliation, granular control, transparency metrics, are coming for them too. That includes search, but also app store summarization, calendar and email AI features, and any third party content aggregation in enterprise AI.
The competing risk for the open web is consolidation in the other direction. Smaller publishers without the legal or analytics resources to evaluate the trade off may default to either always in or always out, neither of which is optimal. Google's metrics will nudge toward always in, while activist boards may push toward always out. The middle ground, granular per template or per section opt outs, only emerges if Google extends the toggle into more fine grained controls, and there is no regulatory mandate yet to do so. We expect a wave of third party tools and consulting offers to fill that gap within ninety days.
What to watch next. The global rollout timeline, which Google said will follow the UK test but did not date specifically. Whether Microsoft Bing and the AI search startups follow with comparable opt outs voluntarily to defuse their own regulatory exposure, or wait to be forced. Whether the EU's AI Act enforcement guidance picks up the CMA pattern for training data. And whether the major US publishers use the new lever in negotiations with Google for licensing renewals due in late 2026. The shape of the open web for the next decade is being negotiated in real time, and today's announcement is the first concrete win publishers have to point to.
A final operator note. For platform engineering teams building internal AI products that ingest web content, the CMA toggle creates a new data quality signal worth tracking. A site that has opted out of AI Overviews has effectively declared that its content is not available for AI ingestion under default terms, and any internal RAG pipeline that crawls the open web should treat that signal as a hard stop, not a hint. We are recommending clients add a check against Google's published opt out list, once it exists, to their data ingestion pipelines, alongside the existing robots.txt and sitemap checks. This is the kind of plumbing that takes one sprint to add and avoids the kind of legal exposure that takes years to unwind.
