Should AI shopping agents face fiduciary duties like human brokers?

Lindsay Owens, writing with Nia Law in an abbreviated version of their Capitol Forum essay, argues that AI agents that shop, invest and negotiate on a person's behalf should be bound by the same fiduciary duties long imposed on human agents such as real estate brokers, lawyers and financial advisors. Owens writes that "autonomous AI agents are operating without the rules, obligations, and guardrails governing their human 'agent' equivalents," and asks plainly, "are they even on your side?" The essay cites Google's new Universal Cart and an updated purchasing protocol, OpenAI's partnership with Plaid for "personalized [financial] guidance," and Robinhood's new autonomous trading and purchasing features as evidence that agentic commerce is scaling faster than any governing rules.

The reasoning draws directly on the common-law doctrine of agency, which Owens describes as centuries old: it "imposes fiduciary responsibilities and conflict-of-interest rules on human agents." The example given is a real estate agent, who cannot represent both buyer and seller in one transaction because "the incentives are irreconcilable." Owens argues the same logic should extend to AI agents built and deployed by companies whose bottom line can conflict with the user's interest — the agent should have to act "in the consumer's interest, not your own." The essay treats fiduciary duty as "a necessary starting point" but "likely insufficient on its own," pairing it with calls for antitrust enforcement and data-privacy protection, since existing consumer-protection and endorsement-disclosure law already gives enforcers some tools.

This sits close to Can traditional governance tools actually control AI agents?, which reaches a similar structural diagnosis — that principal-agent economics explains why AI agents raise new loyalty and monitoring problems — but from an academic, law-review register rather than an advocacy essay aimed at regulators. It also parallels Should AI agents serve one user or balance many interests? in pushing alignment thinking toward accountability structures rather than single-agent loyalty alone, and echoes Can EU consumer law protect AI agents as algorithmic buyers? in treating existing consumer-facing law as inadequate to algorithmic purchasing agents. Where those sources build formal or doctrinal frameworks, Owens' essay grounds the same argument in a rapid sequence of concrete, dated market moves (Google, OpenAI-Plaid, Robinhood) to argue the regulatory window is closing.

The excerpt does not specify what a fiduciary-duty standard would require in enforceable terms, how it would be adjudicated across jurisdictions, or who would bear liability when an agent's recommendation causes harm — it states the principle and the urgency, not the mechanism of enforcement. It also does not establish that current law entirely lacks tools, only that "existing laws... give enforcers a number of existing tools," which the essay treats as underused rather than absent. The implication the essay draws, that policymakers have "a narrow window" before "the infrastructure is hardened," is an urgency claim rather than a demonstrated timeline.