Senator Warner Introduces AI AGENT Act: First Federal Legislation on Agentic AI Systems

Summary

Senator Mark Warner (D-Va.) released a discussion draft of the AI AGENT Act on June 29, 2026, marking the first major federal legislative effort to regulate agentic AI systems. The proposed legislation seeks to protect consumers from platform-controlled AI agents that prioritize corporate interests over user welfare, while promoting competition through independent, consumer-authorized AI agents. The Act defines 'custodial user agents' (CUAs) as software agents that can interact with large platforms on consumers' behalf in transparent, scope-limited, and revocable ways. Using Amazon and its Alexa assistant as a key example, the Act would require platforms with 50+ million U.S. customers to provide interoperable interfaces through which consumers can deploy their own agents to manage purchases, content selection, and account settings.

The legislation establishes critical safeguards for agentic AI governance: agents must have legally cognizable fiduciary duties to consumers; consumer preferences must be legible to agents; agent behavior must be transparent; and interactions must protect privacy and security. By drawing regulatory inspiration from data portability and telephone number portability—frameworks that successfully enabled competition in prior markets—the Act aims to prevent large platforms from deploying 'double agents' that appear to serve users while maximizing platform profit. The discussion draft acknowledges that several prerequisites remain unmet before full implementation, which explains its current status as a consultation document seeking public input before formal legislation.

• The legislation imposes fiduciary duties on AI agents and requires transparent, legible agent behavior to prevent conflicts of interest
• The regulatory approach parallels proven frameworks like data portability and phone number portability that successfully increased market competition