On the morning of June 3, 2026, the UK's Competition and Markets Authority (CMA) delivered exactly what publishers had been demanding for nearly a year: it mandated that Google provide them with genuine control over how their articles are utilized in AI-generated search summaries. The regulator described the move as a "world first," and judging by how rapidly the news spread through the industry, that claim was hardly an exaggeration.
The core of the ruling can be summed up in a single sentence. News websites, magazines, and reference materials will now have the power to block Google from using their content in AI Overviews and AI Mode while maintaining their presence in standard search results. Previously, they were simply never given that choice.
Why the previous model left publishers with no way out
The dilemma was brutal. For Google to see a site at all, a search crawler must be allowed access. However, that same crawler also harvested content for generative features. The only way to opt out of AI summaries was to block indexing entirely, which effectively meant disappearing from search results altogether. In its submission to the regulator, Guardian Media Group explicitly stated that, according to documents from US antitrust proceedings, Google had considered an "opt-out of AI but stay in search" option—and rejected it for commercial rather than technical reasons.
This led to the media's grievance, which the CMA ultimately found valid: AI summaries answer users directly on the results page, preventing them from visiting the source website and causing publishers to lose views, advertising revenue, and bargaining power—even though it is their content that makes the summary useful in the first place.
What exactly Google is now required to do
Legally, the decision is based on the Digital Markets, Competition and Consumers Act 2024. In September 2025, the CMA designated Google as having "strategic market status" in the search sector, which granted the regulator the authority to impose specific requirements. These current rules represent the first of those mandates.
Google is now required to provide clear source attribution with "prominent links" within its AI responses. Publishers are gaining the right to opt out of their content being used—not just in the summaries themselves, but also for fine-tuning models; this option was added following public consultations to ensure the choice covers all scenarios. Every six months during the first year, the company must publish compliance reports backed by specific metrics. While nine months have been allotted for full implementation, the CMA expects the primary tools to be available to publishers well before the deadline.
There is one critical detail, however, without which this entire endeavor would be little more than a hollow gesture: Google has pledged not to use the AI opt-out as a ranking factor in standard search. Publishers had specifically feared a "shadow penalty"—the risk that rejecting AI features would lead to a quiet demotion in search results. During consultations, one industry participant admitted almost verbatim that media companies would not touch these controls without a cast-iron guarantee on this specific point.
Why this matters far beyond the borders of Britain
The scale of the issue is best understood through numbers, and they are alarming for anyone relying on search traffic for survival.
A preprint study released in May, which analyzed approximately 55,000 trending Google queries over 40 days, showed that AI summaries appeared 13.7% of the time. However, when those queries were framed as questions, that share skyrocketed to 64.7%. This indicates that the very searches for which users previously sought in-depth articles are now increasingly being intercepted by instant summaries on the results page. The same research discovered that 11% of the assertions in AI summaries were unsupported by their cited sources, frequently because essential qualifiers were simply left out.
The case of Wikipedia provides a stark illustration of how this impacts traffic, though two distinct figures must be considered. In the fall of 2025, the Wikimedia Foundation reported that human page views had dropped by roughly 8% over the year, tentatively linking the decline to AI and social media. However, an academic study used a more rigorous approach by comparing English articles that featured AI Overviews with the same entries in Hindi, Indonesian, Japanese, and Portuguese, where the feature had not yet rolled out. The net effect was significantly higher, showing a loss of about 15% of English Wikipedia’s daily traffic. Cultural articles saw the steepest decline, while scientific and technical topics were less affected, likely because a brief summary is a better substitute for a simple answer than for a complex one.
This doesn't mean every website will experience the same level of loss. Nevertheless, the numbers clearly explain the fears of publishers: AI search appropriates the answer, leaving the source with little more than a name in a link.
What comes next
From here, everything depends on execution. If Google provides a user-friendly opt-out button, honest metrics in Search Console, and transparent attribution, publishers will gain real leverage in licensing negotiations and perhaps, eventually, direct compensation. Should these tools prove purely cosmetic, the dispute will move back to the courts and other regulators. The European Commission launched its own antitrust investigation into Google's AI practices back in December 2025, meaning the UK precedent will undoubtedly be scrutinized in both Brussels and Washington.
