New State AI Laws + a Senate Bill on Aging

Idaho's Conversational AI Safety Act regulates any public-facing “conversational AI service” that simulates human interaction, including general-purpose tools like ChatGPT and Claude. It requires three things of operators—both developers and deployers: disclosure when users might not realize they're interacting with AI, a crisis protocol that refers users expressing suicidal ideation to mental health support services, and a prohibition on AI providing professional mental care.

It also includes minor-specific provisions: routine AI disclosures, a ban on addictive unpredictable-reward design, and safeguards against sexual content, simulated emotional dependence, and adult-minor romantic role-play. Interestingly, the law does not require age verification; minor provisions kick in when the operator already knows, or is reasonably certain, that it is speaking with a minor.

Tennessee took the narrowest path among the states, regulating only the marketing of AI tools, and nothing else. A person who develops or deploys an AI system can't advertise or represent to the public that the system is a qualified mental health professional. The law borrows the definition of “qualified mental health professional” from Tennessee's mental health code, which covers licensed psychiatrists, psychologists, social workers, and marital and family therapists.

Maine's law reaffirms that psychotherapy is a protected professional activity by prohibiting AI systems from offering services that would constitute professional mental health practice. The law allows clinicians to use AI in clinical practice only for administrative and supplementary support, such as scheduling, billing, logistics, and drafting notes. All AI output must be reviewed by the therapist.

AI may not make independent therapeutic decisions, directly engage in any therapeutic communication with a client, or generate treatment plans without the clinician's review and approval. This law is similar to other laws governing clinicians' use of AI, but it stands out for including narrow exceptions for IRB-approved research and for AI interventions authorized or reimbursed by CMS.

Nebraska's law mirrors Idaho's, regulating conversational AI tools and asking operators to meet three baseline requirements: disclose when a user is talking to AI, adopt a protocol that refers users expressing suicidal ideation or self-harm to crisis services, and refrain from representing that the service provides professional mental or behavioral health care.

As in Idaho, the focus is on protections for minors, which include routine AI disclosures—at the start of a session and every three hours after—a ban on rewards designed to keep minors engaged, and required safeguards against sexual content, simulated emotional dependence, romantic or sexual innuendo, and adult-minor romantic role-play, plus parental account controls for younger users. Like Idaho, the law includes no age-verification requirement, meaning the minor provisions only kick in when the system is aware of or certain that the user is a minor.