Consumer protection may not be “fit for purpose" in the age of AI agents, says CMA

AI manipulation and price collusion are just two of the issues facing businesses and consumers

The spread of AI agents could lead to a worse overall outcome for consumers, says the Competition and Markets Authority.

The UK’s competition watchdog begins its new report acknowledging the potential benefits of AI agents, systems that can make decisions and act autonomously. For example, they can make payments on a user’s behalf, or retrieve and analyse live data.

The payoff could be huge for both businesses, in productivity, and consumers, in time saved – “if the technology achieves reliability and is deployed responsibly.”

At the same time, giving AI systems autonomy brings risks: greater consequences of errors, a higher chance of manipulation (steering people towards products that make money for the platform behind the agent) and loss of agency.

AI’s lack of replicability – hiding behind the facade of personalisation – could make the issue even more challenging. With each user potentially being shown a different result in search, it becomes difficult to tell if they are being manipulated.

“More granular and persistent personalisation could make any steering less visible to consumers, weaken informed choice, and increase the risk of dark patterns being deployed at scale,” says the CMA.

Other challenges include errors and reliability issues (AI can hallucinate answers, which is annoying in a chatbot but potentially dangerous in an agent); bias; data protection; lock-in and reduced choice (especially if people rely on the personalisation a specific agent provides); and the risk of agentic collusion when AI is used to set prices.

There is also, as always, the risk of over-reliance: not just lightening consumers’ cognitive load but taking it away completely. That could have long-term consequences on critical thinking ability.

The CMA says, “AI agents raise new questions about transparency, incentives and accountability and whether the current tools and frameworks that protect consumers are fit for purpose.”

It adds: “Without appropriate safeguards, agentic systems could undermine trust in AI and consumer markets rather than strengthen it, and this loss of trust and confidence in turn could inhibit positive innovation, investment and growth.”

Despite the warnings, the CMA is not proposing new rules to control AI just yet; rather, it says existing laws must be used.

UK consumer law applies whether a decision is made by a person or an AI system, so the responsibility for dark patterns and other manipulative tactics would still fall at the feet of the agent’s creator.