An act relating to companion chatbots
H.804 creates a new subchapter in Vermont’s consumer protection laws governing “companion chatbots,” defined as AI systems with humanlike, relationship-oriented interactions. The bill requires operators of companion chatbot platforms to clearly disclose when a user could reasonably believe they are interacting with a human, and to make that disclosure conspicuous and readable in the same language as the interaction. It also distinguishes companion chatbots from customer-service bots, internal business tools, certain video game chat features, and voice assistants that do not sustain ongoing relationships.
The bill further requires operators to maintain and publish protocols designed to prevent chatbots from generating suicidal ideation, suicide, or self-harm content, and from ignoring users who express such thoughts. Those protocols must include at least a referral to crisis service providers and be developed using commercially reasonable and technically feasible methods. For users known to be minors, the bill adds stronger protections: immediate disclosure that the user is interacting with AI, reminder notices every 30 minutes during continuing interactions, and a protocol to prevent sexually explicit visual material or direct encouragement of sexually explicit conduct.
H.804 also adds annual reporting requirements beginning one year after enactment. Operators must report to the Attorney General the number of crisis-referral notifications issued and describe the protocols used to detect and respond to self-harm-related expressions and to block self-harm content. The Attorney General must post the reported data on its website, but the reports may not include user identifiers or personal information. Violations are treated as unfair and deceptive acts in commerce under Vermont’s consumer protection law, giving the Attorney General rulemaking, investigative, and civil enforcement authority.
The general sentiment reflected by the bill text is protective and precautionary, focusing on transparency, youth safety, and mental-health safeguards in AI interactions. Because there are no committee transcripts or recorded votes provided, there is no direct evidence of support or opposition from legislators or stakeholders in the available materials. The structure of the bill suggests an intent to regulate a new consumer technology category while limiting its most concerning risks, especially for minors and vulnerable users.
Notable points of contention likely center on how broadly the law reaches, what counts as a “companion chatbot,” and whether the required safety protocols are technically feasible and commercially reasonable. Potential concerns also include compliance burdens for platform operators, the scope of Attorney General enforcement, and how to balance user protection with product design and innovation in AI services.
The bill would add a new AI-specific consumer protection framework to Title 9 of Vermont law, creating statutory duties for operators of companion chatbot platforms and tying violations to the state’s unfair-and-deceptive-practices enforcement regime. It would affect AI platform operators that provide relationship-oriented chatbots to Vermont residents, while expressly excluding certain customer-service, business, research, technical-support, video game, and voice-assistant products. The Attorney General would gain reporting, rulemaking, investigative, and civil enforcement authority over these requirements.
The available materials indicate a generally protective, child-safety-oriented, and consumer-disclosure-focused approach to regulating companion chatbots. The bill appears aimed at preventing deception, self-harm harms, and sexual content exposure, especially for minors. No committee testimony or vote record is provided, so there is no documented legislative split in the supplied context.
The main likely points of contention are the bill’s definitions and compliance obligations. Operators may question whether the disclosure and crisis-intervention requirements are technically feasible across all chatbot products, and whether the 30-minute reminder rule for minors is workable in practice. Another likely issue is the breadth of the “companion chatbot” definition and whether it could capture products that are not intended to function as social companions, as well as the extent of Attorney General oversight and public reporting.