CMA examines agentic AI and algorithmic collusion

This renewed focus was kickstarted earlier this month when the CMA opened a new competition law investigation to examine whether hotel chains operating across the UK used a third-party data analytics provider to share competitively sensitive information in breach of UK competition law.

While companies can use various types of data analytics tools and algorithms to help them make commercial decisions and lead to benefits – such as more intense competition, lower costs, and faster changes in prices to better match demand and supply in markets – such practices may also carry competition law risks if this leads to illegal information exchange or collusion.   

This CMA said the new antitrust investigation reflects “the CMA’s wider commitment to ensuring new technologies support fair competition and do not harm consumers”, and has pointed to the CMA’s work explaining “how businesses using algorithms can stay on the right side of the law.”

Commenting on the investigation, Angelique Bret, a competition and consumer law expert at Pinsent Masons, said: “It is well known that sharing competitively sensitive information between rivals, even if this occurs via third parties, can breach UK and EU competition law in the absence of any extenuating circumstances. It will be interesting to see how the CMA investigation develops and whether it may reveal key elements of the conduct, such as the use of algorithmic pricing or AI tools.”

Alex Stratakis, a competition law expert at Pinsent Masons, added: “The use of third-party revenue management software and other similar solutions could represent a material competition law risk – akin to a cartel – when not properly managed. The use of AI further accelerates the potency of such tools and competition compliance needs to be factored in, if not built in. The US authorities and courts have already been grappling with such issues, and the CMA is understandably following suit.”

The CMA has long focused on how algorithms and tech tools can facilitate illegal collusion in breach of competition law. Earlier this month it published its latest thinking on algorithmic pricing, which considered both the potential benefits and risks of such practices and what steps businesses can take to mitigate their exposure. 

The launch of CMA’s competition law investigation was closely followed by the publication of the agency’s latest work on agentic AI from a consumer protection perspective.  The CMA’s new compliance guidance for businesses confirms that consumer law applies regardless of whether a customer deals with a person or an AI agent; that firms are accountable for what those agents do even when supplied by third parties; and that customers should be told when AI is being used and not be misled by design or messaging. Notably, the CMA also recommends that AI agents should be trained to comply with consumer law in the first place.

Under the Digital Markets, Competition and Consumers Act 2024 (DMCC Act) that came into force in April 2025, traders that breach UK consumer protection rules can face fines of up to 10% of their worldwide turnover. Since then, the CMA has been able to use its “direct enforcement” powers to investigate suspected consumer law infringements, order redress and impose penalties without first going to court.

Tadeusz Gielas, a competition and consumer law expert at Pinsent Masons, said the new guidance sets out the CMA’s expectations for how businesses should comply with consumer law obligations when using agentic AI, and provides useful examples, such as when AI agents are used to run marketing campaigns, process refund requests, respond to customer service queries, and provide a service to customers.

He added: “Being the UK’s primary enforcer of competition law – including the new digital markets competition regime – and consumer protection law, the CMA may use a combination of its wide-ranging legal powers to take enforcement action and drive outcomes aligned with its priority areas.”

Alongside the new consumer law guidance, the CMA has also published detailed research explaining how agentic AI could shift consumers from using tools to delegate outcomes to systems that can “sense, decide and act.” Although it notes that agentic AI has the potential to minimise friction and optimise personalised support, it warns that it could also create risks of manipulation, bias, lock‑in and loss of consumer agency – risks that demand reliability, transparency and accountability by design. 

The CMA’s research paper also highlights close links between consumer law and competition law in the area of agentic AI and the risks of agentic behaviour potentially leading to collusive outcomes.

Bret added that businesses would need to review the CMA’s latest publications on agentic AI and algorithmic collusion carefully to ensure they are operating compliantly. "Businesses deploying or procuring AI agents must be aware of the legal compliance risks associated with use of such technology,” she said. “From a consumer protection perspective, they should ensure clear disclosures in AI‑run consumer journeys, maintain human oversight and straightforward escalation routes, and re‑check pricing, promotion and cancellation procedures against the CMA’s compliance guidance under the DMCC Act, including in areas such as pricing transparency and pressure selling.”

From a competition law perspective, if they haven’t already done so, Bret strongly urged businesses “to implement safeguards” for their use of algorithms, data feeds and third‑party tools, to mitigate any risk of unlawful information exchange or the potential for collusive outcomes.

Stratakis added: “Drawing on the CMA’s consumer law guidance, in the competition law context businesses may likewise be ultimately responsible for actions taken by their agentic AI. Training AI agents to be competition law compliant ‘by design’ may help mitigate – but not completely eliminate – the risk of antitrust infringement.”