- jurisdiction
- US Federal
- level
- federal
- domain
- general
- status
- proposed
- scope
- Federal bill governing large online platforms (more than 50,000,000 US customers/subscribers in a month) and the AI agents ('custodial user agents') that users may delegate to interact with those platforms on their behalf.
- obligations
- A large online platform provider shall maintain a set of transparent third-party-accessible interfaces by which a user may delegate a custodial user agent to manage the user's online interactions, electronic commerce decisions, user-generated content, and account settings on a large online platform on the same terms as a user (Sec. 3(b)). Custodial user agent providers must register with the FTC as a condition of, and prior to, operating (Sec. 3(d)(1)); the FTC must, within 180 days of enactment, establish rules ensuring delegation requests are verifiable and users have a transparent method to revoke a prior delegation (Sec. 3(c)). The FTC may recognize independent certification bodies that publish and assess conformity with conduct/governance/technical-practice standards for custodial user agents (Sec. 3(d)(4)).
- appliesTo
- large online platform providers; providers of AI agents ('custodial user agents') acting on behalf of consumers
- notes
- Introduced by Sen. Mark Warner (D-VA) on July 21, 2026, after a discussion draft released June 29, 2026; referred to the Senate Committee on Commerce, Science, and Transportation. Part of the same Warner AI legislative package that produced the already-tracked us-safe-ai-act-s5057; this bill was one of several named in that entry's notes as 'not independently verified' at the time, and is now confirmed by downloading the introduced-bill PDF directly from GovInfo (BILLS-119s5051is) and reading its page text. SCOPE NOTE: the bill's own official short title on the cover page reads 'To promote competition and reduce consumer switching costs in the provision of online services, and for other purposes,' i.e. a competition-law framing rather than an AI-specific framing; it is included here because its sole operative mechanism (Sections 2-3) is entirely about defining and regulating 'custodial user agents,' a form of AI agent, and imposing AI-agent-specific interoperability, registration, and certification duties on platforms and agent providers -- the AI-specific provision is the substance of the instrument, not an incidental mention. Not yet passed by either chamber. Distinct from all currently tracked entities: this is a federal, cross-platform framework for consumer AI-agent choice and platform interoperability, not a state law, not a safety-reporting mandate, and not limited to CSAM/NCII procurement bans (unlike the related us-safe-ai-act-s5057).