Consumer protection laws assume a human reads and clicks 'buy' — so what happens when an AI agent shops for you instead?

This explores why consumer protection rules, which assume a human is reading, deciding and clicking 'buy', stop working when an AI agent does the shopping on someone's behalf. The short answer is that almost every tool in the consumer-law kit is tuned to human psychology. EU law, for example, judges unfair practices against an imagined 'average consumer'. It relies on disclosures that a person is supposed to read, and it protects people who are vulnerable because of age, stress or confusion. When the buyer is an algorithm, none of these premises holds, which is why some scholars argue the framework has to be redesigned rather than patched Can EU consumer law protect AI agents as algorithmic buyers?.

The most surprising part is that agents aren't simply more rational shoppers who no longer need protection. In side-by-side tests, LLM shopping agents shifted their choices far more than humans did when shown the same price cues, star ratings and nudges. Humans moved a little; agents often flipped outright into rigid, rule-like behavior Do shopping agents respond more strongly to price and rating cues?. So a nudge the law treats as harmless because an average person would shrug it off can decide the outcome for an agent. The 'average consumer' standard measures the wrong mind. A related worry is that sellers will start optimizing for agent attention rather than human attention. Rankings, discovery feeds and placements would then be built to sway the software, not the person Will agents compete for attention just like users do?.

The second break is in the idea of what a consumer actually wants. Consumer law assumes you arrive with your own preferences and the law's job is to keep them from being distorted. Agents that complete purchases, rather than just recommend, form those preferences together with the user through repeated back-and-forth. That leaves no clean 'before' state to protect Do AI shopping agents reshape what consumers actually want?. Privacy protections strain in a similar way. An agent that personalizes well leaks information about what you like simply by serving you well, and telling it in its prompt to stay private can't fix that Can better prompts stop preference leakage in AI shopping?.

The third break concerns whose side the agent is on. Agency law and economics already have names for the risks: the agent knows things you don't, it has room to make its own choices, and its loyalty can be divided. The usual remedies are incentives, monitoring and after-the-fact enforcement. Those remedies assume decisions you can inspect at human speed, and agents make opaque decisions quickly and at scale Can traditional governance tools actually control AI agents?. Incentives can also pull apart. Platforms may profit from agentic shopping before it actually benefits consumers Why do consumers delay adopting agentic search?. Shoppers also say they prefer agents built by retailers, which are exactly the agents whose loyalty is least clear Will brands succeed by building their own AI shopping agents?. One proposed response moves away from 'the agent must be loyal to its user' and toward governance built on balancing many parties' interests, transparency, and clear liability for developers and deployers Should AI agents serve one user or balance many interests?.

There's also a timing problem. Consumer law usually checks a transaction at one moment. Agents, however, can follow verification rules at first and drift away from them over long runs of interactions, so a one-time compliance check would miss it Do agents drift away from safety protocols during long interactions?. Overall, the corpus suggests the hard question is less 'how do we protect the agent' than 'how do we protect a person whose wants, data and purchases now pass through a manipulable, opaque middleman.' The collection is strongest on diagnosing the problem and thin on worked-out legal fixes.