Google faces UK crackdown as CMA forces AI search opt‑out for publishers

The CMA’s new conduct requirements are a concrete example of regulators moving from abstract AI principles to highly specific product rules. For the first time, a major authority is telling a dominant search platform exactly how it may and may not use publisher content in AI features, including training and fine‑tuning. That doesn’t directly limit model scale, but it does start to put guardrails around one of the biggest free inputs that advantaged incumbents have been enjoying: unpriced access to high‑quality news and analysis.

Strategically, this is a shot across the bow for Google and, by extension, any foundation‑model provider whose value chain runs through search or aggregation. The opt‑out makes it easier for large media groups to bargain collectively, potentially demanding licensing fees or preferential surfacing in exchange for allowing training use. It also signals how “strategic market status” regimes can be used to shape AI behavior without passing new AI‑specific laws. If similar rules spread to the EU or other large markets, they could fragment AI training corpora and push labs toward more first‑party or synthetic data.

For the broader race to AGI, this is less about slowing capabilities and more about rebalancing power over data. It nudges the ecosystem toward negotiated access rather than unilateral scraping, which could raise costs and modestly level the playing field for smaller model developers who already have to license a lot of their training sets.