UK Watchdog Opens a Case Over Meta’s Publisher Terms for AI
Key Takeaways
- UK CMA Meta publisher terms came under formal investigation on 30 October 2026, when the Competition and Markets Authority opened a Competition Act 1998 case into Meta’s proposed commercial terms for publishers, advertisers, and other businesses.
- The CMA’s concern is that the proposed terms could let Meta collect content from businesses’ own websites and use it for Meta’s own products, including AI features, without those businesses getting a choice about whether or how their content is used.
- No infringement decision has been made: the CMA must first decide whether Meta holds a dominant position in a relevant UK market, and only then whether the proposed terms abuse that position.
- Meta paused the rollout on 28 October 2026, after the terms were first due on 30 October and then moved to 20 November, so the investigation concerns proposed terms that are on hold, not terms currently in force in the form described.
- Meta says there are no grounds for the investigation and that it does not change its terms without consulting publishers and other market participants; the CMA’s initial information-gathering phase runs from October 2026 to January 2027.
Introduction
On 30 October 2026, the United Kingdom’s competition regulator opened the UK CMA Meta publisher terms case that goes to the heart of one of the AI era’s most contested questions: when a platform’s terms let it use other people’s content to build its own products, who gets to say no? The Competition and Markets Authority (CMA) announced a formal investigation under the Competition Act 1998 into Meta’s proposed commercial terms for businesses that use its platforms — terms that, in the regulator’s words, could allow Meta to collect material from companies’ websites and use it for its own products and services, including AI features, without those companies deciding how and for what purposes the material is used.
The case matters far beyond the legal niceties, because it concerns the raw material of AI itself. Publishers, news organisations, and thousands of smaller businesses produce the content that AI systems learn from and answer with. If a platform’s terms can sweep that content into its own products as a condition of doing business, the balance of power between platforms and publishers shifts — and the CMA has now decided that question is worth a formal investigation rather than a quiet negotiation. But the case is also at its very earliest stage, and much of what has been reported about it in the first days is either wrong or premature. This article separates what the CMA actually said from what it has not yet decided.
UK CMA Meta publisher terms: what the CMA is actually investigating
The investigation, listed as a CA98 case with the status Open, names three companies: Meta Platforms, Inc., Facebook UK Limited, and Facebook Ireland Limited. It concerns Meta’s proposed commercial terms for businesses and publishers — the conditions under which businesses use Meta’s platforms and services. The CMA’s stated concern has two parts, and both need to be kept in view.
The first part is about scope. The terms at issue are not limited to content that a business posts on Facebook or Instagram. According to the CMA, they could allow Meta to collect content from businesses’ own websites — the articles, pages, and material that live on a publisher’s own domain — and use that content to develop and provide Meta’s own products and services. AI features are one stated use of that content, not the only one, but they are the use that has drawn the most attention, because AI training and AI answers are where third-party content is most visibly repurposed.
The second part is about choice. The CMA says the terms could permit that use “without those businesses getting a choice on if or how their content is used”. That is the phrase at the centre of the case. A publisher that wants to reach audiences through Meta’s platforms, the argument runs, would have to accept the content terms as a package, with no separate opt-out for the AI use. Whether that amounts to an unfair condition depends on the legal test the CMA must now apply — and that test is where the case could succeed or fail.
Under the Competition Act 1998, the CMA cannot simply declare terms unfair and move on. It must answer two questions in sequence. First, does Meta hold a dominant position in a relevant UK market? Only if it establishes dominance does it then ask the second question: do the proposed terms abuse that position by imposing unfair conditions? The CMA has made no decision on either question. It has not found that Meta is dominant in the relevant market, and it has not found that the terms are abusive. Any report that describes Meta as having been found in breach of competition law on this matter is wrong.
Two further details from the CMA’s announcement help place the case correctly. Meta does not currently hold a strategic market status (SMS) designation under the UK’s newer digital markets competition regime, so this case is being brought under the older Competition Act powers, not under that regime. And the CMA is separately seeking information from Meta about Meta One, a business subscription service, as well as concerns raised by the Federation of Small Businesses about small businesses posting external links on Facebook and Instagram without paying a subscription fee — but those matters fall outside the scope of this investigation, and no decision to take formal action on them has yet been taken. Conflating them with the publisher-terms case would overstate what the CMA has actually opened.
How the case got here: a rollout that stopped before it started
The timeline is one of the most informative parts of this story, because it shows a regulator and a company negotiating in public, in stages, before any formal case existed. Meta originally planned to introduce the revised commercial terms on 30 October 2026. After engagement with the CMA, that date moved to 20 November. Then, on 28 October, Meta announced a pause to the rollout pending consultation with industry groups. The CMA opened its formal investigation two days later, on 30 October — after the rollout had already been put on hold.
That sequence matters for two reasons. First, it means the investigation is about proposed terms, not about a completed change that publishers are already living under. Businesses are not, at the time of writing, subject to the terms in the form the CMA is examining, because Meta paused them. Second, it shows that the CMA’s opening of a case is not the same as a failure of negotiation. The pause and the consultation are still part of the picture, and the CMA has said it will work toward a swift and satisfactory resolution — though it has set no completion date for the investigation.
The complaints that prompted the case came from stakeholders including the News Media Association, which represents UK news publishers, and Movement for an Open Web, a group concerned with the openness of the web. Their involvement is significant because it frames the case as a publisher concern, not just a competition-law abstraction. News publishers depend on Meta’s platforms to reach audiences, and they are also the businesses whose content is most directly at issue. The CMA’s own language reflects that dependence: it describes thousands of UK businesses that rely on Meta’s services to reach customers and audiences, and it quotes its chief executive, Sarah Cardell, as saying that while digital platforms play an important role in helping UK businesses reach millions of customers around the world, Meta must not abuse its position to impose unfair terms.
Meta’s response, as reported by Reuters and the Wall Street Journal, is that there are no grounds for the investigation, and that the company does not change its terms without consulting publishers and other market participants. That position — that consultation has already happened, and that the terms are therefore legitimate — is now the argument the CMA’s investigation will test against the evidence, rather than against press statements.
What happens next, and what it could mean
The CMA’s case page sets out an indicative timetable that runs from October 2026 to January 2027 for its initial phase of information gathering and engagement with Meta and third parties. That is the window in which affected businesses can submit their position to the regulator, and in which the CMA will gather the evidence it needs to answer its two questions. The timetable may change, and the CMA has not promised a decision by any particular date.
After that phase, the CMA has two broad paths. It may close the investigation, if the evidence does not support proceeding. Or it may issue a statement of objections, setting out its concerns in detail and giving Meta an opportunity to respond, before any final decision. A statement of objections is not a final finding either — it is a formal statement of the case Meta must answer. Readers following this story should therefore expect a long process with several distinct stages, not a single verdict in the near future.
For publishers and content businesses, the practical implications are already visible, even before any decision. First, the terms are paused, so the immediate question is not compliance with new terms but participation in the consultation about them — the window the CMA has opened for third parties to make their case. Second, the case sharpens a question every publisher with AI ambitions or AI anxieties now faces: how to distinguish between content a platform may use to display a link or a preview, and content it may use to train or power its own AI products. The CMA’s framing treats those as different things, and its concern is specifically that the proposed terms blur them without giving businesses a choice. Third, related questions are being decided in other forums at the same time — coverage of the case notes that courts in Norway and France are dealing with related issues — so a publisher operating across markets is tracking the same underlying question in several places at once.
For readers, there is a direct connection to a related development from the same week: the shared AI training opt-out signal backed by OpenAI, Google, Microsoft, and Anthropic, which gives publishers a single robots.txt token to signal that their content should not be used for AI training. That opt-out is a technical mechanism, and it depends on AI companies honouring it. The CMA case is about something different but related: whether a platform’s commercial terms can override that kind of choice by making content use a condition of doing business. The two stories together describe the two fronts on which publisher control over AI use of content is now being contested — one technical, one legal — and neither has yet been settled.
What this case is not
It is worth being explicit about what has not happened, because the first days of coverage have already blurred several lines. Meta has not been found to have broken competition law. Meta has not been designated as having strategic market status under the digital markets regime in this case. The proposed terms have not been implemented in the form under investigation, because Meta paused them. And the CMA has not decided whether Meta is dominant in the relevant market — the first question it must answer before it can even reach the question of abuse.
What has happened is narrower, but still significant: the UK’s competition regulator has opened a formal investigation, on the basis of complaints from publishers and open-web groups, into whether a dominant platform’s proposed terms could impose unfair conditions on the businesses that depend on it. That is the beginning of a legal process, not its conclusion, and the next three months of evidence-gathering will determine whether it becomes a landmark case or a closed file. Either outcome will tell publishers something important about how much control they really have over their own content in the AI era.
For context on how AI companies are approaching content and control from the other side, see our explainer on AI Agents Explained: What They Are, How They Work & Uses, and for Meta’s own AI work, see Meta Muse Image: Features, Uses and Limits in 2026.
