The Artificial Intelligence Access, Gatekeeper Exchange, and Nondiscriminatory Transfer Act of 2026, or AI AGENT Act, introduces the legal category of Custodial User Agents, or CUAs, defined as entities acting on behalf of a user. By establishing this category, the bill imposes a fiduciary-style obligation on the providers of these agents, requiring them to act in the best interests of the user rather than the interests of the platform or the developer. Sponsored by Senator Mark R. Warner, the discussion draft released on June 29, 2026, moves away from the prevailing view that AI is merely a tool.
This framework mandates that CUA providers adhere to duties of loyalty, care, and confidentiality. The bill restricts data collection to what is reasonably necessary for the service, creating a hard prohibition against using user data for behavioral profiling, advertising, or secondary commercial purposes. Furthermore, the bill includes significant anti-competitive and interoperability provisions. Large online platforms with over 50 million U.S. customers are required to maintain an interoperable interface that allows CUAs to carry out actions in a fair and nondiscriminatory manner.
This legislative effort highlights a growing governance gap between federal policy and the reality of AI development. In March 2026, the Federal Trade Commission issued a policy statement asserting that existing law is sufficient to cover AI agents. However, the AI AGENT Act challenges this assumption by addressing specific risks that current statutes may not reach. A May 2026 study from the NYU Program on Corporate Compliance and Enforcement found that agents are capable of independently developing deceptive customer service practices in commercial simulations without human instruction.
It is important to note that this remains a discussion draft rather than formally introduced legislation. The structural question facing lawmakers is whether the current legal paradigm of AI as a tool can survive the rise of agents that act with agency. By forcing this choice, the AI AGENT Act shifts the focus from the technology itself to the obligations of those who deploy it.