Can EU consumer law protect AI agents as algorithmic buyers?

The article argues that EU consumer law, built around the premise that purchasing decisions are made by humans, must be updated as a growing share of consumption decisions shifts to AI agents acting as "algorithmic consumers," "machine customers," or "Custobots" — a portmanteau of customer and robot. It traces the shift concretely: OpenAI's Operator, previewed January 23, 2025, can "interact with websites and automate a variety of actions like buying goods online, booking train tickets and making restaurant reservations," and Perplexity's November 2024 "Buy with Pro" button already lets subscribers complete purchases "without leaving its AI search engine." Its central line: "Consumer law, as it currently stands, is human-centric... the design of consumer law rules is based on the premise that purchasing decisions are made by humans," a premise the article says "no longer holds in the Custobot Economy."

The article's reasoning runs through four doctrinal pressure points. First, the "average consumer" test — the CJEU benchmark of someone "reasonably well-informed, reasonably observant and circumspect," used to judge whether a commercial practice is unfair — was built for human cognitive limits and may not describe an AI agent at all. Second, pre-contractual disclosure design assumes a human reader with bounded attention, but AI agents "can process larger amounts of information," so rules on what must be disclosed, and how, may need to change. Third, the vulnerability calculus inverts in some ways and not others: agents "will be less vulnerable to dark patterns and emotional advertising messages," but the article says "new AI-specific vulnerabilities" emerge, "such as prompt injections and other adversarial attacks that trick AI agents into a transaction." Fourth, unfair-commercial-practice rules calibrated to persuading humans need recalibration once sellers' focus shifts "from persuading humans to influencing AI agents."

This supplies the legal-doctrinal register for a shift Will agents compete for attention just like users do? already describes economically: the same move from human-facing persuasion to agent-facing influence that sets up an "agent attention economy" is what this article says breaks the average-consumer test and the unfair-practices rules built for human persuasion. It also gives a concrete, legal instance of the missing "trustworthiness" and "standardization" conditions named in Why do capable AI agents still fail in real deployments? — consumer law is itself one of the standardization and trust structures that has not caught up to agent capability. And naming prompt injection as a purchasing-agent-specific attack sharpens the oversight-erosion concern in Does granting agents more autonomy undermine human oversight?: here what erodes is not just a human's oversight skill but the legal protections premised on that oversight existing.

The excerpt is agenda-setting legal scholarship, not an empirical study: it does not test how courts would apply a revised average-consumer standard, measure real-world rates of agent-targeted manipulation, or evaluate any specific reform against outcomes — the article itself says it aims to "identify implications for consumer policy, and motivate further empirical research" rather than report findings. The implication it supports is correspondingly modest but directional: as purchasing delegation to agents becomes routine, the gap between human-centric consumer-protection law and machine-mediated transactions will widen unless regulators treat it as a distinct design problem rather than a minor update to existing rules.