Warner proposes federal registry for trusted AI agents and fiduciary rules

Mark Warner unveiled a discussion draft on June 28 for the Artificial Intelligence Access, Gatekeeper Exchange, and Nondiscriminatory Transfer Act, a bill that would create a federal registry for trusted AI agents and impose security and accountability rules on software that can act with little human oversight.

The draft defines an AI agent as a system that can perceive its environment, act autonomously or under preestablished rules or objectives, and make decisions with minimal human oversight in service of a defined goal. Warner’s proposal says those systems are already being used for research assistance, cybersecurity, software development, scheduling, email management, customer service and support, which puts the fight over agentic AI squarely in the consumer-protection lane.

Warner’s plan would set Federal Trade Commission security standards and create a federally vetted list of AI agent software providers. It is also framed around human ownership and secure use of agents on social media and other online platforms, a sign that the senator sees the next round of AI abuse as less about flashy chat and more about software that can click, buy, sign up and negotiate on a user’s behalf.

The sharpest legal test in the draft is Warner’s push to make AI agents in financial services owe a duty of loyalty to the principal they represent, like fiduciaries. That idea would try to answer a basic question now looming over autonomous software: if an agent can move money, open accounts or steer a transaction, who is legally responsible when its incentives drift away from the consumer’s interests?

Warner has already been pressing the U.S. Department of the Treasury on agentic AI, asking questions four weeks before June 30 about rulemaking, existing laws and security risks. The senator’s broader 2026 AI agenda also includes the Workforce Transparency Act and the AI Labeling Act, which places the registry proposal inside a wider federal effort to force more disclosure, more oversight and more accountability across the AI stack.

The timing matters because the FTC has already moved against AI-related consumer harms. On September 25, 2024, the agency launched Operation AI Comply, a sweep of five enforcement actions aimed at deceptive or unfair conduct tied to artificial intelligence. On September 11, 2025, it opened an inquiry into consumer-facing AI chatbots, seeking information on how companies measure, test and monitor potentially negative impacts. Warner’s proposal now asks whether a federal registry and fiduciary rules would genuinely protect consumers, or simply bless bots that still fail in opaque ways once they are handed the power to act.