The UK could force one of the most consequential changes yet to the default-search market: Google may have to present qualifying AI assistants alongside conventional search engines when people choose their default search service on Android and Chrome.
In strengthened proposals published on September 23, the UK Competition and Markets Authority said Google should give users a choice of search service when they first use an Android phone or open Chrome, followed by another prompt every year. The regulator explicitly says AI assistants should be eligible when they meet the relevant criteria, citing the growing use of services such as ChatGPT and Perplexity as alternatives to traditional web search.
There is a second condition with major implications for publishers. Search providers appearing on the choice screens would need to attribute publisher content fairly, allowing users to understand where information came from and easily access the original source. The proposal therefore links search distribution to source attribution: an AI assistant could gain access to one of the web’s most valuable default-selection surfaces, but only under rules intended to preserve a visible route back to the content it uses.
The CMA’s official announcement opens consultation on the strengthened proposal until October 9, 2026. A final decision is expected by the end of the year. Nothing has been imposed yet, and ChatGPT or Perplexity should not be described as guaranteed choices; they are examples of the category of AI assistants the CMA wants the rules to accommodate.
The search choice screen could stop being a list of search engines
Traditional default-search regulation assumes that the relevant competitors are search engines. The CMA is updating that assumption because user behavior is changing faster than the regulatory category.
The regulator says people in the UK increasingly use AI assistants as search services, yet those assistants are not included on Google’s current choice screens. Its proposal would make the eligibility framework broad enough for AI assistants to participate when they satisfy technical, security and other applicable requirements.
That changes the competitive unit. The choice would no longer necessarily be Google versus another conventional search engine with a familiar list of ranked links. It could become Google versus a conversational assistant that searches, synthesizes information and returns an answer with citations.
For users, those are fundamentally different interfaces for accessing the web. For publishers and marketers, they are fundamentally different discovery systems.
The CMA wants the choice at setup—and then again every year
The proposed intervention is designed to make the decision recurring rather than permanent. Google would have to show a search-service choice when a customer first uses an Android phone or opens Chrome, then prompt the user annually to select the default provider they want.
That annual prompt matters because default positions are powerful partly through inertia. A user may select a service once and continue using it for years without revisiting the decision, even as competing products improve or entirely new categories emerge.
A recurring choice screen would periodically reopen distribution. An AI assistant that was immature or unavailable when a device was purchased could become an eligible alternative later.
In effect, the CMA is proposing a recurring competition event on two of the most important gateways to web search.
ChatGPT and Perplexity are examples, not guaranteed winners
The CMA’s announcement explicitly names ChatGPT and Perplexity when describing the rise of AI assistants, but the proposal does not award either service a place on the screen.
Eligibility would depend on the final rules and on providers meeting the relevant technical and security requirements. The regulator is consulting precisely because the detailed design still needs to be settled.
This distinction is important. The headline development is not that the UK has chosen ChatGPT or Perplexity as defaults. It is that the regulator wants the default-search category to become technologically neutral enough that an AI assistant can compete for that position.
If adopted, the framework could potentially accommodate services that do not yet exist, provided they satisfy the criteria.
Source attribution becomes part of market access
The publisher requirement may be the most distinctive part of the proposal. The CMA says any search provider featured on the choice screens should fairly attribute publisher content when it is used, so people can easily access the source and understand where search results originated.
That turns attribution from a product-design preference into a possible condition of access to regulated distribution.
An AI assistant could not simply qualify because it can answer questions well. If it relies on editorial content, the way it exposes those sources to users would matter to its eligibility under the proposed framework.
For publishers, this creates a direct connection between competition policy and AI citation design. The question is no longer only whether an AI company chooses to provide visible sources. Source accessibility could become part of the rules governing whether that service gets access to Google-controlled choice screens.
Attribution is not the same as traffic
Even a strong attribution rule would not guarantee publisher clicks. An AI assistant can clearly identify a source while still satisfying the user inside the generated answer.
The practical impact will depend on what “fairly attribute” means in the final requirement: how prominent the source is, whether it is directly clickable, how citations map to claims and whether the original publisher remains understandable in a synthesized answer.
This distinction matters because generative search changes the relationship between visibility and referral traffic. A publisher can be used as evidence and receive a citation without receiving the same visit that a traditional search result might have generated.
NetContentSEO recently covered a randomized field experiment in which forced use of Google AI Mode reduced external click-through by 18.8 percentage points compared with ordinary Google Search. One study cannot establish the effect of every AI-search product, but it illustrates why attribution and traffic should be measured separately.
The proposal targets Google because defaults are strategic infrastructure
The CMA designated Google as having strategic market status in general search and search advertising services in October 2025. That designation allows the regulator to impose targeted conduct requirements under the UK’s digital markets regime.
The regulator’s case is built around the importance of access points. Android and Chrome can determine which search service is placed in front of millions of users before competitors have a chance to win them through product quality alone.
The CMA has previously cited evidence that default positions are commercially valuable and can create barriers to entry and expansion. The strengthened proposal attempts to reduce that advantage by making the user choose and by periodically reopening the choice.
Adding AI assistants recognizes that the competitive threat to Google Search may increasingly come from products that do not look like search engines at all.
Default AI search could reshape the economics of GEO
If conversational assistants can become default search services, generative engine optimization moves closer to the center of mainstream search strategy.
Today, many organizations treat ChatGPT, Perplexity and other AI systems as secondary discovery channels alongside Google. A regulated default-choice mechanism could change that distribution equation by giving qualifying assistants direct access to users at browser and device setup.
A user who chooses an AI assistant as the default search service may begin far more information journeys inside a synthesized-answer environment. That would affect which sources are retrieved, how brands are represented and where publisher links appear.
The SEO question would no longer be simply how to rank in Google while monitoring AI citations on the side. Businesses could be optimizing for multiple competing default discovery systems.
Different default providers can create different source markets
AI search systems do not necessarily retrieve or cite the same sources for the same question. Their indexes, retrieval systems, ranking logic, tool use and model behavior can produce materially different evidence pools.
NetContentSEO has documented research showing that ChatGPT, Gemini, Copilot and Perplexity can expose different citation-quality risks, including the use of AI-generated source material. Other studies have likewise found substantial variation in source selection across generative systems.
That means a default-choice screen could do more than redistribute search volume among brands. It could redistribute visibility among publishers.
A domain that performs strongly in Google may not enjoy the same prominence in Perplexity or ChatGPT, while a source that rarely ranks in conventional search could become important inside an assistant’s retrieval pipeline.
The CMA is effectively future-proofing the definition of search
One of the most important aspects of the proposal is conceptual. Regulation often arrives after technology has already changed the market definition on which the rules were built.
The CMA is trying to avoid that lag by defining the choice requirement around the function users want—finding information—rather than locking it to the interface conventions of a traditional search engine.
Its announcement says the strengthened proposal responds to rapid changes in how people search and specifically notes that AI assistants are increasingly used as search services.
If the final rule retains that approach, future AI products could compete for default status without first having to resemble Google Search.
Google would still control important implementation details, but under regulatory constraints
A choice screen sounds simple, but its design can strongly influence user behavior. Ordering, descriptions, logos, default states, explanation text and the number of providers displayed can all affect selection.
The CMA’s broader user-choice work is therefore about “choice architecture,” not merely the existence of a settings menu. The regulator wants the decision presented in a way that gives competing services a meaningful opportunity to be selected.
The final conduct requirement will need to translate that principle into implementation rules. AI assistants complicate the task because a user choosing between Google and ChatGPT is not comparing two interchangeable interfaces.
The screen may need to communicate what each service actually does without steering the user toward one business model or interaction style.
Publishers could gain leverage from competition between answer engines
If several AI and search providers compete for default status under a common attribution requirement, source treatment can become part of product competition.
One provider might expose citations prominently and make original reporting easy to open. Another might provide minimal attribution while still technically satisfying the rules. Publishers, regulators and users would then have more concrete grounds for comparing how these systems treat source material.
This could create pressure beyond the minimum legal requirement. A search assistant that wants to position itself as trustworthy may decide that transparent sourcing is a competitive feature rather than merely a compliance obligation.
The CMA proposal therefore has the potential to influence interface design even outside the UK if providers standardize their citation systems across markets.
The UK is connecting user choice with publisher protections
The strengthened choice proposal is not occurring in isolation. The CMA has already imposed other conduct requirements on Google under the digital markets regime.
In June 2026, the regulator imposed publisher controls intended to give publishers greater influence over how their content is used in Google’s AI features. It subsequently imposed fair-ranking and data-portability requirements.
The new proposal adds user choice to that framework while extending publisher considerations to competing search providers that want access to the choice screen.
The result is a regulatory model that treats the search market as a three-sided system: users need meaningful choice, competing providers need access to distribution, and publishers need protections around how their content is used and attributed.
The consultation could still change the details
The proposal remains open for comment until October 9. The CMA says it will consider responses before reaching a final decision, which it expects by the end of 2026.
That means the eligibility criteria, presentation rules and attribution obligations could change before implementation. Google, AI providers, publishers, consumer groups and other interested parties can still challenge or refine the proposed approach.
Any analysis should therefore distinguish the regulator’s current direction from an enacted requirement. Android and Chrome users in the UK are not yet being shown ChatGPT or Perplexity as regulated default-search choices under this proposal.
What has changed is that the regulator now explicitly wants the rules to make that outcome possible.
Default search is becoming a contest between answer architectures
For two decades, the default-search battle largely meant competition over which search engine received a query from the browser address bar. Generative AI is breaking that model.
The UK’s proposal recognizes that a user can now begin the same information journey with an AI assistant, receive a synthesized answer, inspect citations and continue through a conversation without ever experiencing a conventional search-results page.
If the CMA’s proposal becomes a final requirement, Google could be required to place those different architectures in front of users as genuine alternatives at setup and again every year. Qualifying AI assistants would gain a route into default distribution, while publisher attribution would become part of the price of admission.
That combination is what makes the proposal significant for SEO and GEO. The competition is no longer only for rankings inside one search engine. It is increasingly a competition to become the search layer itself—and the UK is considering rules that could let users choose which layer they want.