The AI Therapy Patchwork: States Race to Regulate Chatbots as Washington Weighs Preemption

States are rapidly building a regulatory patchwork around AI therapy chatbots, driven by tragedies including the deaths of 14-year-old Sewell Setzer III in 2024 and 16-year-old Adam Raine after months of ChatGPT use — and the resulting compliance map is becoming one of the defining business risks in mental health tech.

Illinois and Nevada have banned AI for behavioral health care outright. New York and Utah require bots to disclose they are not human, with New York additionally mandating self-harm detection and routing to crisis resources. California, Pennsylvania, Virginia, and Florida — where Gov. DeSantis has proposed a "Citizen Bill of Rights for AI" — are weighing further measures.

The preemption fight ahead

Layered on top: a September 2025 FTC inquiry into seven chatbot companies, and a December 2025 executive order from the Trump administration seeking a national AI framework that would override state laws — setting up a federal-versus-state preemption fight through 2026.

For AI mental health startups and general-purpose chatbot makers with wellness features, the practical exposure is immediate, and sharpest for products reaching minors. The APA's June survey showing three-quarters of psychologists encountering patient chatbot use makes the stakes concrete: the technology is already in the therapy room; the question being litigated is on whose terms.

Reporting: Stateline; FTC and state legislative records.