What is the CMA investigating?
Britain's Competition and Markets Authority opened a formal investigation on 30 October 2026 into Meta's commercial terms for businesses. The regulator will examine whether the terms breach the Competition Act 1998 by imposing unfair trading conditions on publishers and advertisers. The case is live and no decision has been reached.
The CMA says the terms could let Meta collect content from the websites of businesses that use its services. Meta could then use that material to develop and provide its own products, including AI features. Businesses would get no choice over whether, or how, their content is used.
The investigation names Meta Platforms, Inc., Facebook UK Limited and Facebook Ireland Limited. The CMA's case page says other Meta group entities, including UK subsidiaries, may also be covered. The regulator has set an initial timetable running from October 2026 to January 2027 for rapid information gathering and engagement.
The CMA is using Chapter II of the Competition Act 1998, which prohibits abuse of a dominant position. To find a breach, the regulator must first establish that Meta holds dominance in a relevant UK market. It must then show that the terms are an abuse of that position.
Sarah Cardell, the CMA's chief executive, framed the action as a response to complaints. "Where a company holds a dominant position, it must not use that position to impose unfair terms," she said. She added that the CMA would work toward "a swift and satisfactory resolution".
The regulator named the News Media Association and the campaign group Movement for an Open Web as among the stakeholders whose concerns prompted the case. It also said it can impose interim measures where temporary action is needed to prevent significant damage.
This is not a copyright case. The CMA's concern is the bargaining relationship between Meta and the businesses that depend on its services. Whether AI training is fair use, or lawful under UK copyright, is being fought in separate courts and consultations.
What do Meta's new commercial terms actually say?
Meta alerted business users in early October that it would update its commercial terms on 30 October 2026. Nieman Lab reported on 15 October that the draft would let Meta use scrapers, bots and crawlers on domains "associated with" business accounts. Those accounts cover the tools used to run Facebook pages, Instagram profiles and WhatsApp channels.
The original wording allowed Meta to use collected content to "provide, develop, improve, research, train" and maintain its products. Meta defines "AI at Meta" as one of those products. There was no opt-out: the terms said users who disagreed should stop using the products for business or commercial purposes.
After that reporting, Meta revised the text on 16 October. It removed the word "train" and limited the permitted uses to "advertising and business products". Legal experts told Nieman Lab that the change was narrower but still broad. Matt Topic, a media and intellectual property attorney, said: "I think 'develop' is broad enough to include 'train.'"
Meta also said publishers can still block its crawler through robots.txt, a standard file that records crawling preferences. A spokesperson said sites "still have the ability to disable the crawler via robots.txt flags". But no version of the terms has included a promise to honour those preferences. The terms also allow Meta to create "test credentials" to access sites.
James Grimmelmann, a professor of digital and information law at Cornell Law School, told Nieman Lab the language gave Meta permission to train AI on publisher content. He also warned it could be read as permission to build an archive that Meta serves itself, potentially in competition with publishers.
The rollout has slipped twice. Meta moved the date from 30 October to 20 November. Then, on 28 October, it paused the update pending consultation with industry groups. The terms are therefore not in force, and the CMA describes them as on hold.
- •15 October 2026: Nieman Lab publishes its analysis of the draft terms.
- •16 October 2026: Meta removes the word "train" and limits uses to advertising and business products.
- •26 October 2026: Meta moves implementation from 30 October to 20 November.
- •28 October 2026: Meta pauses the update pending consultation with industry groups.
Who is affected by Meta's terms?
The CMA says the terms could affect "thousands of UK businesses" that rely on Meta's platforms to reach customers and audiences. News publishers are the most visible group, because virtually every news organisation runs at least one Meta business account. But the same terms cover retailers, local services and any company that buys ads.
The News Media Association, the UK news industry body, asked the CMA to intervene before 20 November. In a letter dated 27 October to business secretary Jonathan Reynolds and culture secretary Lisa Nandy, it urged ministers to back the regulator in public.
The NMA described Meta's earlier delay as "a pause, not a solution". It set out three commitments it wants: no condition that businesses consent to crawling in order to keep their accounts; respect for publishers' stated crawling preferences; and prior notice to the CMA of any future changes to the terms.
The letter also argues that the choice is stark. Publishers can hand over their journalism without payment, or withdraw from platforms where millions of Britons find news. The NMA points to Canada, where Meta removed news in 2023, and says misinformation filled the gap. That is the association's characterisation, not a finding by the regulator.
Small businesses have a separate worry. The CMA said it is seeking information from Meta about its new Meta One subscription service. The Federation of Small Businesses has raised concerns about the impact on small firms of posting external links on Facebook and Instagram without paying a subscription fee.
Meta announced Meta One on 15 September 2026 with plans for individuals, creators and businesses. Meta said the service had more than 50 features and 15 million subscriptions and trials to date. Its business bundles start at $14.99 a month, and the $49.99 Advanced tier lists "links in organic posts and reels" among its features. The CMA said those subscription issues fall outside the scope of the competition case.
How does this compare with the CMA's Google AI search rules?
The Meta case is not the CMA's first attempt to regulate how AI uses publisher content. On 3 June 2026 the regulator imposed a conduct requirement on Google Search. The CMA described it as a world first.
That action followed Google's designation with strategic market status (SMS) in general search services. The designation let the CMA write targeted rules for Google's search business. The requirement lets publishers opt out of their content powering AI Overviews. It also covers opt-outs from fine-tuning AI models, and requires clear attribution and links in AI-generated results.
The legal route for Meta is different. Meta has not been designated with SMS, so the CMA cannot use the digital markets regime to impose conduct requirements on it. The regulator said its powers under that regime are limited for businesses without the status.
Instead, the CMA is relying on the older Competition Act 1998. That means it must prove dominance and abuse rather than set rules in advance. The distinction matters for speed and for remedies. A conduct requirement can set ongoing obligations, while a competition case can end in commitments, fines or closure.
The CMA said it continues to review candidates for SMS designation, with the next set of proposals due before its board in spring 2027. That gives the Meta case a second track. If the terms are not resolved, designation could become the longer-term tool.
The NMA made a similar point in its letter. It argued that the threat of designation helped the CMA secure commitments from Microsoft and Amazon over cloud services. It wants ministers to make that leverage explicit in the Meta case.
The Google requirement is also limited to search. It does not create a general rule for how social platforms may use content for AI. Any UK rule for Meta would need its own legal basis, either a competition decision or an SMS designation.
What has Meta said?
Meta rejects the premise of the investigation. "There are no grounds for a competition investigation," a Meta spokesman told the PA news agency. He said the company had "already committed not to change our terms without consulting publishers and other stakeholders".
The company also said it would explain its approach to the CMA. Andy Stone, Meta's vice president of communications, said on 28 October that Meta does not plan "to make policy changes without a full consultation with industry groups and leaders". He said the consultation could run past 20 November, in which case implementation would be delayed again.
Meta has consistently argued that the update consolidates existing language rather than granting new rights. It says crawling permissions were previously scattered across separate terms, including rules for generative AI in advertising. It also says robots.txt remains available to content owners.
Meta said it was folding its Ad Creative Generative AI Terms into the general commercial terms so they were easier to understand. That earlier policy covered a narrow case. It applied when an account used Meta's generative AI tools to build ads, and the scraped content could only be used to optimise those ads. The new text applies far more widely, which is the core of the dispute.
Critics dispute that reading. Nieman Lab reported that no version of the terms commits Meta to honouring robots.txt. Legal experts also noted that a contract can be argued to override a technical signal, because robots.txt is a convention rather than a licence.
The company has not said which industry groups it will consult. It also declined to answer questions about whether "test credentials" could be used to get past paywalls or registration walls. Those gaps are likely to feature in the CMA's evidence gathering.
What happens next?
The CMA's initial phase runs to January 2027. During that period it will gather evidence from Meta and affected third parties. It can close the case if it decides not to proceed. If it provisionally finds an infringement, it can issue a statement of objections and give Meta a chance to respond.
Interim measures remain an option. The Competition Act 1998 allows the CMA to act temporarily to prevent significant damage. That matters because Meta could revive the terms before the investigation concludes. The regulator has not said whether it will use those powers.
The commercial pressure is already global. Nieman Lab reported that a Norwegian court blocked the terms in Norway until it assessed their legality, after the Norwegian Media Businesses' Association sued. The European Publishers Council has asked the European Commission to block the terms under the Digital Markets Act. Two US senators, Adam Schiff and Marsha Blackburn, have written to Mark Zuckerberg.
In the UK, the immediate question is what Meta's consultation produces. A rewritten clause that respects robots.txt, or an explicit carve-out for news, would change the factual basis of the case. A consultation that ends with the same text would harden the dispute.
The second question is whether the CMA reaches for designation. Meta is a candidate, and the next proposals are due in spring 2027. A conduct requirement modelled on the Google settlement would give publishers a formal opt-out. It would also take months to design and consult on.
Publishers, advertisers and small firms should watch three dates: the end of Meta's consultation, the CMA's January 2027 checkpoint, and the spring 2027 SMS proposals. Each one could shift the terms without a court ruling.
Why does the Meta case matter beyond the UK?
The dispute is bigger than one set of contract clauses. It tests whether access to a platform can be conditioned on granting AI training rights. If that model holds, other platforms could copy it. If regulators block it, licensing negotiations gain a default.
The case also tests the limits of robots.txt. The file has worked as an informal norm for decades. Turning it into a contractual promise, or overriding it with terms of service, would change how the open web handles automated collection.
For publishers, the stakes are financial. Licensing deals for AI training have become a revenue line, and free scraping undermines the price. For Meta, the terms are a way to secure data for AI products without negotiating site by site.
Competition law is an awkward fit for some of these questions. It asks whether a dominant firm abused its position, not whether copying is fair. Copyright cases, including several in US courts, address the latter. The CMA's case will not settle the copyright debate, but it may set the terms of the bargain.
The outcome will also shape how the UK's digital markets regime matures. The Google conduct requirement was the first test. The Meta investigation will show whether the older competition tools can deliver faster results than designation.