An act relating to regulating the use of artificial intelligence in the provision of mental health services
H.816 regulates how artificial intelligence may be used in the provision of mental health services in Vermont. The bill states that therapeutic judgment, clinical decisionmaking, and therapeutic communication must remain the responsibility of qualified mental health professionals, while allowing AI to be used for administrative, operational, documentation, and quality-improvement tasks. It defines key terms such as artificial intelligence, generative artificial intelligence, mental health services, therapeutic communication, clinical responsibility, and consent, and it distinguishes between permitted support functions and prohibited clinical functions.
The bill prohibits any person or entity from offering, providing, or advertising mental health services as though AI itself provides diagnosis, treatment, therapeutic judgment, or therapeutic communication. It also bars mental health professionals from using AI to make therapeutic decisions, directly interact with clients, generate treatment plans, or detect emotions or mental states. At the same time, it allows AI-assisted transcription, recordkeeping, scheduling, billing, claims processing, de-identified data analysis, and workforce planning, so long as the professional retains clinical responsibility and, in some cases, obtains patient consent. The bill applies consumer-protection remedies to violations and treats misuse by mental health professionals as unprofessional conduct for licensing purposes.
H.816 would amend Vermont professional-regulation and consumer-protection law by adding a new chapter governing artificial intelligence in regulated professions and by making misuse of AI in mental health practice a disciplinary offense. It would also amend the unprofessional-conduct provisions for relevant licensing boards so that violations can trigger professional discipline, and it would make prohibited AI-based mental health offerings enforceable under the Consumer Protection Act, giving the Attorney General and private parties civil remedies. The bill further directs the Artificial Intelligence Advisory Council to study the issue and report recommendations to legislative committees by January 15, 2027.
The bill’s stated purpose and structure reflect a generally cautious, protective approach toward AI in mental health care. Its findings and definitions emphasize patient safety, human clinical oversight, informed consent, and the preservation of non-clinical uses of AI that may improve efficiency and access. The inclusion of exceptions for religious counseling, peer support, and self-help resources suggests an effort to avoid overregulating nontraditional or nonclinical services while still drawing a firm line around licensed therapeutic practice.
The main point of contention is the scope of the prohibition on AI in mental health services, especially whether AI should be allowed to participate in therapeutic communication, diagnosis, treatment planning, or emotion detection. The bill takes a restrictive position, but it also preserves room for AI-assisted administrative and supplementary functions, which may be seen as a compromise between innovation and safety. Another likely area of debate is the breadth of the bill’s coverage of licensed, certified, rostered, and even some unlicensed providers, as well as whether the consent and disclosure requirements are workable in practice for providers using AI-enabled tools.