Google’s AI opt-out leaves publishers with a choice they can’t safely use
Publishers have finally won the right to keep their articles out of Google’s AI search results. The harder question now is whether that right is usable when the system keeps them opted in by default, and Google withholds the AI-specific data they’d need to judge if pulling out is worth the risk.
The U.K.’s Competition and Markets Authority has ordered Google to give publishers greater control and transparency over how their content is used in AI search, including an opt-out from AI Overviews, AI Mode, and AI summaries in Discover, clearer attribution, and improved reporting on how content appears in AI-generated results. Publishers will also be able to opt out of AI model training.
The ruling, handed down last week, is the first time a major regulator has forced Google to give publishers a formal say over how their work is fed into AI features in search.
Google issued its own blog the same day as the CMA announcement, saying that while the CMA ruling applies to U.K. sites, it will roll out the opt-out toggle in its search console globally. It also reiterated the recent measures it’s taken to reintroduce publisher-friendly features like inlinks and Preferred Sources within AI Overviews and AI Mode.
On paper, it looks like a long‑sought victory for an industry worried about losing traffic, attribution and control as Google leans further into AI‑generated answers. But with nine months to implement changes, Google has left publishers wondering how much protection this really offers in a zero-click world.
That delay, combined with the fact it’s an opt-out rather than an opt-in system – and that Google is drip-feeding the data they’d need to assess the impact – leaves publishers with a theoretical right to say no to AI in search that is, in practice, almost unusable, stressed Stuart Forrest, former global audience development director for Bauer Media Group. “If this were opt-in, Google would have to prove to us why we should let them use our content,” he said. “Instead, they keep us in by default and don’t give us the click data we need to know if AI search is actually hurting our business.”
But without AI‑specific click data, publishers can’t see whether AI Overviews are actually sending traffic, or just answering questions on the page and killing clicks, noted Forrest. “They’ve broken out search impressions about AI features, but they haven’t given us search clicks for AI features – that doesn’t exist,” he said.
The practical questions now are who, if anyone, will flip the toggle and what level of traffic loss makes that untenable. Many believe the combination of opt‑out, lack of data and competitive pressure means very few publishers will actually pull their content.
While trade bodies and some publishers have publicly welcomed the ruling, though behind closed doors, there are major concerns. Paul Bannister, chief strategy officer at Raptive, called the opt-out option “nonsensical” for publishers, adding that he doubted anyone would use it. “Giving publishers a front-end toggle for AI Overviews or AI Mode is not real control; it is Google offering publishers a light switch while keeping the power plant running,” he said.
Publishers aren’t asking for a “cosmetic” switch that only controls where Google shows AI answers, stressed Bannister. They want two stronger protections: a guaranteed way to stop their content from being used to train Google’s AI models; and a way to block Google’s AI systems from pulling information from their sites when generating responses. “Google is once again answering the question it wishes publishers had asked, not the one they actually asked,” he added. “It will delay, obfuscate and claim it is doing the right thing while carefully avoiding the thing publishers are actually asking for.”
Google did not respond to a request for comment in time for this article’s publication.
It remains “impossible” for publishers to judge the consequences of opting out versus staying in, so many publishers may decide to stay because “they fear the unknown,” said Thomas Lue Lytzen, director of ad sales and tech at Danish news publisher Ekstra Bladet. “Publishers still lack granular control over how their content is used for AI, meaningful insight into the traffic being generated, and assurance that users will actually be directed back to the original source rather than consuming publisher content inside Google’s ecosystem,” he said.
Another big problem is that the opt-out lets Google group multiple services, like AI Mode and AI Overviews, together, rather than letting publishers control each one separately, stressed Chris Dicker, CEO of Candr Media Group. “My concern is that Google could eventually bundle Discover into the same framework, knowing full well that many publishers now rely on Discover as a major source of traffic and therefore cannot realistically opt out,” he said.
Many publishers are still furious that Google’s product decisions are choking off traffic, even as demand for their content rises elsewhere and they deploy non-impression-based strategies.
Case in point: recent earnings from People Inc., a large publicly-listed, multi-brand publisher, showed that Google sessions in Q1 of 2024 through to Q1 of 2026 resulted in a total loss of 800 million visits. People Inc. is large and diversified enough to absorb that kind of hit: many smaller or less diversified publishers simply wouldn’t be.
The coming months will test how real the choice actually is. Some subscription‑heavy and niche outlets may be tempted to experiment with opting out, betting that protecting perceived value matters more than marginal search referrals. Ad‑dependent publishers, by contrast, have far less room to move: any meaningful hit to Google traffic could blow a hole in their economics.
Publishers like Reuters will treat this as a risk calculation: is Google’s AI search actually delivering enough value to justify participating? And without proper reporting on how many clicks come from AI Overviews and AI Mode, they have no reliable way to answer that question.
“There’s not enough data,” said Phil Andraos, general manager of Reuters Digital. “If there was, we could make that decision very quickly. Instead, it’s going to be a more careful evaluation, probably testing as well.”
The nine-month timeline seems glacial to most, given the speed at which AI is reshaping search. Jason Kint, CEO of publisher association Digital Content Next, argued that the remedy only touches future use of publisher content in AI search, not “the vast amount of protected content already forcefully taken and used to train AI models without permission or compensation.”
While the CMA ruling is the clearest regulatory intervention so far, Kint sees it as only one front in a broader legal fight over Google’s AI practices. “U.S. federal courts have already ruled Google has an illegal monopoly in search, so injunctive relief to end the harm to the open web from not having a real choice to opt out of training is on the front burner,” he said, pointing to the ongoing Penske vs Google case.
Forrest fears hardly anyone will use the opt-out and that Google will weaponize that. He believes a remedy that almost no publisher dares to use is a gift to Google: the company can point to low opt‑out adoption as proof that publishers are happy with AI search, even if the real reason they stay in is a lack of data and fear of losing ground to competitors.
He also fears the U.K. agreement could become the template other regulators copy. If this first deal is built on an opt-out system and incomplete data, that weaker standard could be imported into Brussels – the seat of the European Commission – and used to argue there’s no need for tougher rules, he added.
The CMA’s executive director for digital markets Will Hayter, said in a Press Gazette interview, that the “opt-in vs opt-out” debate is being overblown. “It’s up to Google to decide to put forward how they plan to comply with the control, and that could, in principle, be opt-in or opt-out,” he said.
Several supply-side ad tech industry execs pointed to the new attribution provision as a positive. But publishers still don’t think it’s enough to replace the click. “A link buried inside, or beneath, a 500-word AI-generated answer that effectively replaces the original article is worth far less than attribution attached to a short search snippet that encouraged users to click through,” said Dicker.
A head of SEO at a large lifestyle publisher described the provision of AI impression data as “a start,” but added that Google’s reluctance to offer AI clickthrough data is likely because the numbers would show clicks are low. “It would contradict what Google has been saying about how AIOs are good for publishers… Impressions is the positive spin on AIO for Google. Clicks would prove publishers’ point that it is hurting traffic.”
But even that potential upside sits uncomfortably alongside where Google is clearly trying to take search next. At Google I/O last month, the direction of travel was hard to miss. In a closed-door briefing with Google search execs and around 80 large publishers, representing what one exec in attendance described as “a billion dollars of content investment,” the tech giant faced some uncomfortable questions. It was pressed on why publishers should keep creating content for a platform that now wants to be “the destination rather than the highway” – a tension that sits squarely at the heart of the U.K. CMA’s proposed rules on how Google can use publisher content to power AI in search.
They didn’t get a clear answer.
“I/O was the clearest example that Google treats the internet like their own fiefdom, and they assume the rights to use all content on the internet because they assume the internet is Google’s – that’s a lie,” said the exec.
Big publishers have moved from grumbling to actively scenario‑planning: some are blocking Google Extended and have already opted out of Gemini, others are modelling de-indexing from Google. “The kind of thing that looks suicidal historically is something that pretty much everyone’s got a model for,” said the same exec.
Set against that backdrop, the CMA’s opt‑out starts to look less like a clean win and more like a high‑risk trade‑off: publishers can tell Google not to feed their content into AI Overviews, but doing so may mean stepping back from the very surfaces where Google now expects users to spend most of their time. And that’s a choice many publishers still don’t feel they can afford to make.
A CMA spokesperson said the new conduct requirement imposed by the CMA strengthens the bargaining power of publishers with Google, giving them control over how and when their content is used in AI features.
“We’ll actively monitor how Google will implement these changes – including assessing the implications for businesses,” said the spokesperson in an email statement to Digiday. “If needed, the CMA will bring forward work on further measures to ensure a fair exchange of value takes place between Google and publishers.”
Seb Joseph contributed reporting.
More in Media
YouTube’s AI remix push exposes a looming reckoning for the creator economy
YouTube’s Gemini Omni integration has highlighted some of the major problems generative AI poses in the creator economy.
Why creator Lola Torres prefers the stability of affiliate marketing over brand partnerships
Creator Lola Torres on the hustle of building her career in affiliate marketing, the challenge of creator programs, and more.
Media Briefing: Perplexity’s new ‘trust and transparency’ pitch does little to win over publishers
Perplexity wants to be a trusted partner to publishers, but a growing list of copyright lawsuits are making that a difficult sell.