UK Choice Screens Gate AI Assistant Defaults

The Competition and Markets Authority closes consultation at 5pm on 9 October 2026 on an updated user-choice conduct requirement for Google’s general search services. Under the proposal, Android and Chrome users in the UK would face a search-service choice at first use, then an annual prompt to set a default. Qualifying AI assistants could appear alongside conventional search providers.

That is the dominant story. After designating Google with strategic market status in general search and search advertising in October 2025, the CMA has already imposed publisher controls, fair-ranking, and data-portability requirements. The 23 September package strengthens the choice-architecture piece: technical and security criteria for assistants, plus a requirement that providers on the screens attribute publisher content fairly so users can see where answers came from. A final decision on this conduct requirement is expected by the end of the year.

If first-use and annual choice prompts are expected to weaken Google’s default-distribution advantage, why is there still no public evidence that users will switch providers—or remain with an alternative after making a selection?

Eligibility comes before the prompt

The screen is not an open bazaar. Assistants must meet relevant criteria, including technical and security requirements, before they can compete for exposure. That sequence matters: the distribution fight starts after qualification, not at the moment a user taps a tile. Eligibility, review procedures, and any challenge rights in the final remedy text will decide which products ever reach the scarce real estate of first-use and annual prompts.

Publisher attribution tightens the same gate. The CMA’s release frames fair attribution as a condition for featuring on the screens—people should reach source content and know where results came from. That rule is not a footnote. It can shrink the set of assistants that may treat publisher material as free fuel while claiming default distribution. Europe’s DMA debate has already shown how search data and chatbot access can diverge under platform rules; Britain is writing a related bargain into choice architecture.

Selection is not retention

The consultation page also lists a dedicated “User research on choice screens in search” study. That document is the nearest official evidence object. It does not, by itself, publish a UK series of selection rates, one-year retention, or repeat-prompt behavior after an assistant is set as default. Without those numbers, markets that price “AI on the choice screen” as automatic parity with Google Search are pricing a slot, not a sticky habit.

Default gravity, inattention, and product quality can unwind an initial switch. That is the same utility gap problem in consumer form: distribution access does not prove habitual use. A roundup of European competition cases can note that Britain wants assistants on search-choice screens; it cannot invent durable switching data that the consultation record has not yet produced.

What still looks mis-priced is treating screen inclusion as the end of Google’s default advantage. The unresolved variables are who qualifies, how the screen is ordered and designed, and whether year-one selections survive habit and product comparison. China’s efficiency narrative and Silicon Valley’s buildout bets reprice capability; this UK rule may reprice distribution—but only if retention follows selection.

The claim to watch is narrow. The piece holds until the CMA’s final remedy text specifies qualification tests and until any public implementation data—or a credible pilot—shows durable redistribution after first-use and annual prompts. Representative user data showing substantial, lasting switches away from Google, or a pilot where annual prompts produce persistent usage shifts rather than one-time clicks, would kill the residual. Until then, Britain’s choice screen is a gate under design, not a proven reallocation of search demand.