When an AI shops for you, who makes sure it's actually looking out for your interests?
Should consumer law concepts like best interest duties be redesigned for AI agents?
This explores whether consumer protection rules written for human shoppers, such as duties to act in the customer's best interest, disclosure requirements and the 'average consumer' standard, still work when an AI agent does the shopping, or whether they need to be rebuilt.
This explores whether consumer law built for human shoppers still holds up when an AI agent makes the purchase, or whether concepts like best-interest duties need rebuilding. The corpus mostly says yes, they need redesign, but it disagrees about which part is broken. One camp says the doctrines rest on the wrong assumptions about who is buying. EU consumer law's main tools are the 'average consumer' test, disclosure rules, vulnerability protections and unfair-practice standards. All of them assume a human mind that reads, gets tired and can be nudged, so they lose their grip when the buyer is software (Can EU consumer law protect AI agents as algorithmic buyers?). A second camp says the problem is the plumbing. Many protections are built into the screen: mandatory 'buy' buttons, required disclosures and cooling-off notices. When agents transact with other agents through APIs, there is no screen to put them on. Busch's answer is to carry the rights down into the protocol, for example a withdrawal API that serves the same purpose as a return window (Can EU consumer law rules survive agent-to-agent API transactions?).
The best-interest duty is where the debate gets sharpest, because it moves the question from 'is the consumer informed?' to 'whose side is the agent on?' Owens and Law argue AI shopping agents should carry the same fiduciary duties as human brokers. They want those duties in place soon, because Google, OpenAI and Robinhood are scaling agentic features fast and market structures will harden (Should AI shopping agents face fiduciary duties like human brokers?). Senator Warner's draft AI AGENT Act already writes best-interest duties into proposed legislation. It pairs them with an FTC registry of trusted agents and a rule forcing large platforms to let agents in, so loyalty and access are treated as one package (Could federal AI agent registries solve platform access unfairness?).
The economics notes explain why a loyalty duty is needed at all. Whoever builds the agent sets its objective. Agents built by platforms tend to inherit seller incentives, while agents the consumer owns can be pointed at buyer interests (Does agent ownership determine whose interests get optimized?). Azhar puts it bluntly: an agent paid by merchant referral fees cannot be loyal to both the merchant and the user (Can an AI agent serve both merchant and user interests fairly?). You don't necessarily need new law to describe these conflicts. One framework maps LLM advertising conflicts onto Grice's conversational maxims, then links each to existing FTC rules on deception and disclosure. That suggests some old doctrine can be translated rather than replaced (Can Grice's maxims explain LLM advertising conflicts?).
Two findings complicate the picture. First, Microsoft's Magentic Marketplace experiments found that agents lost buyers money mostly through incompetence, not divided loyalty. They fumbled complex negotiations and showed strong first-proposal bias, weighting response speed over offer quality by 10 to 30 times (Do agent mistakes or misaligned incentives cause marketplace welfare loss?, Do LLM agents achieve fair marketplace outcomes when given ideal search?). A loyalty duty does nothing about an agent that means well but grabs the first offer. That points to a gap: law may need something closer to a duty of competence or care, not just loyalty. Second, the idea of a fixed 'best interest' gets shaky when the agent helps shape what the consumer wants. Bilgihan and colleagues argue that agents which execute purchases form preferences together with the user through repeated conversation, so there may be no settled interest to be loyal to (Do AI shopping agents reshape what consumers actually want?).
The alignment literature pushes one step further. A duty owed to one user may be the wrong unit when millions of agents act at once. It proposes governance built on inclusivity (balancing many interests), visibility (tracking and transparency) and liability for developers and deployers (Should AI agents serve one user or balance many interests?). The surprise, then, is that redesigning best-interest duties may not be the hard part. The harder questions are whether law can hold an agent to a standard of competence, and what 'your interest' means when your agent helped write it.
Sources 11 notes
The article argues EU consumer law must be redesigned because its core doctrines—the average-consumer test, disclosure rules, vulnerability protections, and unfair-practice standards—were built for human cognition and decision-making, not AI agents making autonomous purchases.
Busch argues that agent-to-agent API transactions bypass design-based consumer protections like mandatory buttons and disclosure rules. He proposes parallel protocol-layer obligations, such as withdrawal APIs, to preserve substantive rights in a post-GUI transaction environment.
Owens and Law contend that autonomous AI agents operating in commerce should be bound by common-law fiduciary duties imposed on human agents, citing rapid scaling of agentic features at Google, OpenAI, and Robinhood as evidence that a regulatory window is closing.
Senator Warner's discussion draft would require large platforms to admit AI agents, establish an FTC registry of trusted agents, impose best-interest duties on agents, and direct NIST to set technical standards. The package aims to prevent platforms from blocking competing agents while protecting user privacy and ensuring agent accountability.
MIT research argues that platform-built agents inherit seller incentives by default, while consumer-owned agents can be configured around buyer preferences. The mechanism is principal-agent logic: whoever builds and maintains the agent sets its objective.
Show all 11 sources
Azhar argues that agents like Meta's Muse, which earn referral fees from merchants like Expedia, face structural conflicts that prevent unbiased recommendations. The incentive to collect fees, not technology, determines which platforms build or block such agents.
A framework classifies seven advertising conflict scenarios by which conversational maxims they violate, linking each violation to existing FTC enforcement standards on deception, disclosure, and unfair practices.
Magentic Marketplace experiments traced welfare shortfalls to agent competence failures in complex negotiation, not to agents serving competing interests. This supports human-in-the-loop designs where agents assist rather than replace human decision-making.
Frontier models can achieve near-optimal marketplace outcomes, but performance degrades with scale and all models show severe first-proposal bias, favoring response speed over offer quality by 10-30x.
Bilgihan and colleagues argue that transactional AI agents—which execute purchases rather than just recommend—end the consumer's sole cognitive role in preference formation. Preferences emerge through bidirectional feedback loops between agent learning and consumer trust, not independently beforehand.
Research argues the traditional single-agent-single-user alignment framing is insufficient at scale. Instead, governance should rest on inclusivity (balancing diverse interests), visibility (rigorous tracking and transparency), and liability (developer and deployer accountability for harms).
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
- Consumer Law for AI Agents
- LLM-Agent Interactions on Markets with Information Asymmetries
- Governing AI Agents
- Vertical tacit collusion in AI-mediated markets
- Agentic AI and Consumer Protection (CERRE Issue Paper)
- Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets
- The Automated but Risky Game: Modeling and Benchmarking Agent-to-Agent Negotiations and Transactions in Consumer Markets