Amendment to H.816 would require licensed clinician review of AI-generated mental-health output before delivery

A proposed amendment to H.816 reviewed by the Senate Health & Welfare committee would narrow the bill's language about AI in mental-health services and add a professional-review requirement.

The amendment removes a reference to "an individual"—language that judiciary worried could be interpreted to subject parents or informal caregivers to enforcement—and instead requires that "mental health services shall only be provided in Vermont through AI if a mental health professional reviews and approves the AI-generated output prior to the information being transmitted to the patient, client, or AI user." The presenter said this was intended to target chatbots rather than informal interpersonal caregiving.

The change would require that if an entity hosts a clinically used chatbot in Vermont, a mental-health professional licensed in Vermont would need to review AI output before delivery to users. The presenter acknowledged implementation challenges—for example, how hosting entities would contract or hire licensed clinicians to vet outputs—and said judiciary had concerns that are being addressed in editor language.

Committee members said they would review the precise amendment language with staff (Katie) before the bill goes to the floor; the amendment is expected to be finalized and coordinated with judiciary before floor action the following day.