Britain’s competition regulator has stopped consulting and started ordering. On Wednesday the Competition and Markets Authority imposed new conduct requirements on Google’s search services, the first concrete obligations to follow from its decision to designate the company as holding strategic market status.
Among them is a provision with sharp implications for the AI era: publishers will be able to opt out of having their content train Google’s AI models.
Google’s search results increasingly summarise the web rather than send users to it, and those summaries are built on content the company crawls for ranking. Publishers have argued they face a trap: refuse the crawl and vanish from search, allow it and feed the AI systems that reduce their traffic. The CMA’s rule is meant to break that bind by separating the two, letting a site appear in search without consenting to AI training.
The rest of the package is structural. The CMA’s requirements mandate fair ranking, transparency, proper content attribution, and default choice screens on Android and Chrome so users can pick rival search services rather than accept Google’s by default. Choice screens are familiar from a decade of EU antitrust enforcement; their inclusion here signals the CMA intends to use the practical levers that have moved the needle elsewhere.
The legal architecture is what makes this different from a one-off case. Under the UK’s Digital Markets, Competition and Consumers regime, the CMA can designate a firm with strategic market status in a digital activity and then impose tailored, ongoing conduct requirements, rather than litigating each abuse separately.
The CMA confirmed Google’s SMS designation in search in October 2025, opened a consultation on conduct requirements in January 2026, and has now moved to enforcement.
It is a different model from the American one. Where US antitrust against Google runs through the courts, with remedies argued case by case over years, the UK regime is regulatory and forward-looking: a designated firm operates under a standing set of obligations a regulator can adjust. The trade-off is familiar.
The regulatory approach is faster and more flexible; it also concentrates a great deal of discretion in the regulator, and Brussels has tended to write the rule before settling what enforcement actually looks like in practice.
For Google, the requirements arrive as its search business is already under pressure from a direction antitrust was not designed for. AI assistants and chat interfaces are emerging as alternatives to the ten blue links, and the CMA’s rules implicitly acknowledge that the competitive threat and the regulatory question now both run through AI. The opt-out provision in particular is less about classic search rivalry than about who controls the training data underneath the next interface.
Google has consistently argued that its services benefit users and that heavy-handed rules risk degrading them, a line it will presumably repeat here. The company has room to challenge specifics, and the detail of compliance, how the opt-out works technically, how choice screens are designed, tends to be where these regimes are won and lost.
What is settled is the direction. The UK has built a standing regulatory relationship with Google’s search business and has now used it for the first time. The immediate question is whether an AI-training opt-out can be made to work without pushing publishers out of search by other means. The CMA has written the rule. Enforcement is the part still being drafted.
Stephan is the sports journalist for the Maple Grove Report.
