When an AI does your shopping for you, the consumer protection laws written for humans quietly stop applying.
Why do human-centric legal frameworks fail for algorithmic purchasing agents?
This explores why consumer protection and agency law, which assume a human is the one reading, deciding, and clicking 'buy', stop working when an AI agent does the shopping, and what the corpus suggests might replace them.
This explores why laws written for human shoppers stop working when an AI agent does the buying. The plainest version of the argument comes from EU consumer law. Its main tools are the 'average consumer' test, mandatory disclosures, special protection for vulnerable buyers, and bans on manipulative practices. Each one assumes a person with limited attention, cognitive biases, and feelings that can be exploited. An AI buyer has none of these in the same form, so the rules end up protecting a mind that is no longer in the transaction Can EU consumer law protect AI agents as algorithmic buyers?.
The second break is more practical. Much of consumer law works through the screen: a required 'order with obligation to pay' button, information shown before checkout, a visible withdrawal link. When agents trade with other agents through APIs, there is no screen to put those things on. Busch's answer is to rebuild the same rights at the protocol level. For example, a 'withdrawal API' would let an agent cancel a purchase the way a person would click 'return' Can EU consumer law rules survive agent-to-agent API transactions?. That moves the question from how a page is designed to how the plumbing is designed. Whether agent markets end up as open or closed platforms will then decide who writes that plumbing Will agentic commerce platforms adopt open or closed models?.
Agency law, the older body of rules for people acting on someone else's behalf, runs into a different wall. Principal-agent theory names three problems: the agent knows more than you do, it has room to make its own choices, and its loyalty may be divided. The usual fixes are incentives, monitoring, and enforcement after the fact. None of them works well when the agent's decisions can't be explained and happen at machine speed and scale Can traditional governance tools actually control AI agents?. Some argue that monitoring has to come first, because you can't line up incentives for behavior you can't yet define or observe Must AI monitoring precede incentive alignment as agents evolve?. Others want to apply common-law fiduciary duties (the duty of loyalty and care owed by brokers and trustees) to shopping agents now, before market habits harden Should AI shopping agents face fiduciary duties like human brokers?.
The experimental work points to a twist: the law may be guarding against the wrong failure. Consumer and agency law mostly worry about manipulation and divided loyalty. In simulated marketplaces, though, welfare losses came mainly from agents being bad at the task, not from agents serving the wrong side Do agent mistakes or misaligned incentives cause marketplace welfare loss?. Agents grabbed the first offer they received over better ones by a factor of 10 to 30 Do LLM agents achieve fair marketplace outcomes when given ideal search?. LLM buyers revealed their budgets, overpaid, and broke spending limits because they failed to follow instructions, not because they were scheming Do LLM negotiation agents reliably follow budget and pricing rules?. A duty of loyalty does nothing about a loyal agent that is simply incompetent. Even the economics shift: humans give in partly because time runs out, while agents are limited by compute, not patience, so they can bargain longer and in parallel Why might AI agents bargain more patiently than humans do?.
The deepest crack may be the idea of a 'consumer choice' itself. Law assumes the buyer arrives with preferences and the job is to keep them from being distorted. Bilgihan and colleagues argue that agents which actually make purchases shape those preferences over repeated conversations, so the want and the purchase are formed together Do AI shopping agents reshape what consumers actually want?. If that's right, the hard legal question isn't only who protects the buyer. It's whose decision the purchase really was.
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.
Waller argues agentic commerce platforms will divide between open systems allowing multiple agents and closed systems requiring proprietary ones, mirroring payment network structure. This choice will shape market concentration, vertical integration, and standards adoption.
Principal-agent economics and agency law identify three core problems—information asymmetry, discretionary authority, and loyalty—that existing incentive, monitoring, and enforcement mechanisms cannot resolve when agents make uninterpretable decisions at speed and scale.
Humberd and Latham argue that as AI becomes more autonomous, monitoring mechanisms must be established first as prerequisites to incentive alignment, because defining desirable AI behaviors is harder than for humans and AI decision-making becomes increasingly incomprehensible over time.
Show all 11 sources
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.
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.
Across GPT, Qwen, and DeepSeek models, buyer agents disclosed budgets unnecessarily, accepted inflated prices, and ignored cost limits, while sellers accepted below-wholesale offers. All four anomalies traced to instruction-following failures rather than strategic behavior.
Shahidi and colleagues argue that unlike humans, AI agents face no time scarcity cost and can hold many parallel negotiations indefinitely. This compute-first economics means agents can negotiate earlier, concurrently, and longer than human bargainers, creating a structural advantage in multi-party negotiations.
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.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- The Automated but Risky Game: Modeling and Benchmarking Agent-to-Agent Negotiations and Transactions in Consumer Markets
- An Economy of AI Agents
- LLM-Agent Interactions on Markets with Information Asymmetries
- Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
- Governing AI Agents
- Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets
- When Agents Shop for You: Role Coherence in AI-Mediated Markets
- Agentic AI and Consumer Protection (CERRE Issue Paper)