March 10, 2026

UK Regulator Reminds Businesses: Consumer Law Still Applies to AI Agents

By James Stephen

The UK’s Competition and Markets Authority (CMA) has published new guidance making clear that consumer protection law applies in full when businesses deploy AI agents, and that companies will be held responsible for what those agents do.

The CMA assures it is not looking to slow AI adoption. It points, instead, to the UK being the third largest artificial intelligence market in the world, after the US and China, and says it is “committed to encouraging” the use of AI to help boost the country’s economic growth and improve people’s everyday lives. This enthusiasm comes with a clear message, however, that AI does not create a legal grey zone.

UK consumer protection law has long applied to businesses’ use of technology, and the CMA has clarified that it is the “effects on consumer decision-making, not the novelty of the technology, that matter most”. The Digital Markets, Competition and Consumers Act 2024 has since strengthened that framework, making clear that businesses must not mislead, manipulate or pressure consumers regardless of how those outcomes are produced.

What Businesses Need to Know

In guidance published on GOV.UK, the CMA sets out four core obligations for any business using AI agents in customer-facing roles.

• Transparency: Businesses must tell customers when they are interacting with an AI agent rather than a person, particularly where that distinction might affect the customer’s decisions. Overstating what AI can do, or obscuring its role, risks breaching consumer law.
• Training: Businesses are expected to think carefully about how their AI agents are configured: whether they respect statutory rights and cancellation terms, whether they could mislead customers through what they say or omit, and whether they properly handle consent. Testing is also a “crucial” part of training AI agents and failing to take the appropriate training measures could mean breaking the law.
• Monitoring: Once an AI agent is live, businesses must check regularly that it is delivering the right results and behaving as intended. The CMA highlights the risk of AI “hallucination”, where a model generates inaccurate or nonsensical outputs, and says human oversight is essential to catch errors before they cause harm.
• Rapid response: If a problem is identified, businesses must act quickly to address it, especially where an agent is interacting with large numbers of customers or with vulnerable people.

Critically, the CMA confirms that responsibility cannot be passed to a third-party AI provider. If a business procures an AI agent from elsewhere, it is still accountable for that agent’s conduct and could face fines of up to 10 percent of global turnover if things go wrong.

Agentic AI: The Bigger Picture

Many routine consumer tasks, including comparing offers, switching providers and managing subscriptions, are time-consuming and mentally demanding, and these burdens fall disproportionately on households with less time or confidence to engage. We are already seeing technology companies pivot towards agentic commerce technology, for example. Agentic AI could handle a range of tasks continuously in the background, such as securing better deals and reducing friction without repeated manual effort. Gartner also predicts that agentic AI will autonomously resolve 80 percent of customer service by 2029. At scale, the CMA suggests, AI could meaningfully improve consumer outcomes and household prosperity.

The risks, however, are significant. The CMA flags the possibility that commercial pressures could lead to manipulative design, or “dark patterns”, being deployed at scale through highly personalised systems. Added to this, there are concerns about errors having real financial consequences, AI amplifying bias in decision-making, and consumers becoming over-reliant on automated systems. The regulator also flags risks around autonomous pricing agents damping competition, and challenges around data privacy and accountability.

The outcomes for businesses and consumers are intrinsically connected as well, particularly considering consumers are less forgiving of AI mistakes than human ones, according to a recent Medallia report. Managing these risks well, the CMA suggests, is therefore what will allow agentic AI to become a positive asset: “The UK has an opportunity to position itself at the forefront of trusted agentic innovation, fostering a dynamic, competitive ecosystem that drives household prosperity, innovation, and growth.”