More AI Could Be Coming to Google Search Under Upcoming UK Regulations

Trade associations in the UK have warned the country’s competition regulator that proposed rules for Google may harm publishers by injecting even more AI into search.

The Competition and Markets Authority (CMA) proposed a variety of conduct requirements for Google earlier this year, some of which, like a rule letting publishers opt out of AI Overviews, have been cautiously hailed as wins for the industry.

But another requirement, giving users greater choice over their default search engine in Google’s platforms, has raised alarm because it is worded in a way that could see consumers offered the choice to automatically use AI services like ChatGPT and Perplexity for search.

Many publishers have seen massive traffic declines since Google began rolling out AI Overviews in search. The fear is that, if consumers were given the option of easily making ChatGPT their default search engine on Google’s Chrome browser or an Android phone, that pattern could accelerate further.

At present, UK users of Android phones (except Google Pixels) are asked once, when setting up their phone, whether they would like to use a search engine besides Google as their default. This “choice screen” is not shown for users of the Google Chrome browser on desktops or iOS devices, who automatically get Google as their default search engine.

Deeming this a competition issue, the CMA’s proposed conduct requirements would compel Google to give users of many of its platforms a choice screen once a year, as well as an optional 14-day “test-drive” period at the end of which users are asked whether they want to stick with their chosen search engine.

At the moment, Google’s own choice screens do not offer users the option to use an AI service as their default search engine. The CMA wrote in its proposals that this may be a problem, and that eligibility criteria for inclusion on the choice screen should be widened.

“The search sector is rapidly changing and the way users search the web is evolving, including through the growth of new generative AI-based services,” the regulator wrote.

“We want to ensure that the eligibility criteria for inclusion in choice screens appropriately include relevant competing providers of search services—particularly as offerings, and user expectations and behaviour, evolve.”

Eligibility for inclusion in the Android choice screen is currently assessed by Google on an annual basis, but the CMA said it wants this timeline to halve “given the pace at which search services are evolving, potentially meaning the Search Choice Screen fails to reflect the latest developments in the market.”

‘Serious Concerns’

The proposed conduct requirements were published in January, setting off a consultation period that has now ended. Although the CMA has not yet published its final decision on the user choice requirements, responses to the consultation have been made public—and reveal worries among publishers and their representatives.

Trade association the News Media Association said it had “serious concerns” about the possible inclusion of AI assistants in the choice screen.

“Given the very low referral rates associated with AI assistant interfaces, this could accelerate traffic diversion from AI search services where publishers have an ability to opt-out towards bad actors who refuse to recognise publisher technical controls, including robots.txt, [and] have significant adverse consequences for publishers’ revenues and sustainability.”

It added that would “sit uneasily” with the goals of the CMA’s other conduct requirements ordering Google to, for example, give publishers greater control over the use of their content in its AI surfaces.

Another industry body, the PPA (Professional Publishers Association), similarly said that “the inclusion of AI assistant tools within the choice architecture would raise distinct concerns and risk undermining the purpose of the remedy…

“AI chatbots such as ChatGPT and Perplexity are materially different from search engines. They are not designed primarily to direct users to external websites, but instead present information directly, often synthesized from third party content, including copyrighted material, without necessarily driving traffic to rightsholders or operating within established licensing frameworks.”

One point that was raised repeatedly was that, when setting out the scope for these conduct requirements last year, the CMA for now chose to exclude Google’s AI service Gemini on the basis that evidence it is being used by consumers as a general search engine is limited.

“While Gemini remains outside the scope of the [strategic market status] designation, it would be inconsistent and harmful to treat those same services as equivalent to general search engines,” said the NMA.

The Guardian’s parent company agreed, writing that it was “inconsistent to define other AI assistants as equivalent search engines for the purpose of choice screens” when Gemini was excluded.

One anonymous respondent said the CMA should consider requiring separate choice screens for traditional search engines and for AI ones.

“Even though LLMs have begun to redefine how consumers research online, they remain, for now and for the foreseeable future, distinct products with different use cases and search intent,” they wrote, noting evidence that one-third of searches are navigational—i.e. someone typing in a brand name rather than asking a question.

AI Companies Voice Support For Google Choice Screen Inclusion

The impact of whatever the CMA chooses to do may not be limited to the UK: after the regulator made its AI Overviews opt-out decision, Google said it would ultimately roll out the capability globally.

And indeed the U.S.’ News/Media Alliance weighed in, telling the CMA that “the inclusion of AI assistants in a choice screen would risk undermining the other content requirements.” It argued that, quite apart from offering users more opportunities to use AI, there should instead be a “clear and easy way for users to opt-out of or disable generative AI experiences in Google Search.”

Several U.S. tech companies also made submissions to the consultation. Microsoft welcomed the widened eligibility criteria, noting explicitly that they were “broad enough to encompass AI chat interfaces, which Microsoft believes to be an important and growing source of general search queries.”

ChatGPT creator OpenAI went further, saying the CMA should design the conduct requirement to be “sufficiently forward looking and broad enough to capture how people’s search habits are evolving, including search that incorporates AI.”

It went on: “It would be a bad outcome if Google can point to innovat[ive] AI chatbots as their most capable competitors and substitutes for general search services – as well as have Google’s own search service prominently use an AI interface and AI models – while also excluding these innovate AI chatbots that have search functionality from a choice screen which facilitates consumer choice.”

For its part, Google responded to the consultation arguing that imposing more choice screens would be “interruptive” and “frustrating” for consumers, but said it was willing to use the choice screen eligibility criteria stipulated by the CMA.

The CMA has not indicated when it plans to publish its final conduct requirements, nor which way it’s likely to go on including AI services in the choice screen.

In a document summarizing a roundtable on the matter, the regulator said a publisher trade association had asked if AI services were likely to qualify for the choice screen: the CMA said “that it wished to use this consultation to collect views on this issue and did not wish to prejudge which services might be included at this stage.”

The post appeared first on A Media Operator.