MONTPELIER, Vt. — Vermont has enacted legislation regulating the use of artificial intelligence in mental health services, effectively prohibiting independent AI therapy bots from providing therapeutic services in the state.
The measure, House bill H.816, was delivered to Gov. Phil Scott on June 11 after winning final approval from both chambers of the Legislature, according to legislative tracking records. It amends state law to classify certain uses of AI in mental health care as violations of the Consumer Protection Act.
The legislation prohibits the offering or advertising of mental health services that rely on AI to deliver therapeutic communications, clinical guidance, diagnoses, treatment plans or other therapeutic decisions, unless provided by a qualified licensed mental health professional or as part of an approved research study. Violations by licensed providers would also constitute unprofessional conduct.
“Our primary intent was to prohibit independent AI therapy bots, whether they explicitly or implicitly are providing therapeutic services, from being able to do so in Vermont,” Lynn Currier, executive director of the Vermont chapter of the National Association of Social Workers, said in a May interview with VTDigger.
The bill advanced amid growing concerns about the capabilities and risks of AI chatbots in sensitive areas like mental health. Supporters argued it ensures that therapeutic judgments remain with licensed human professionals. It allows AI for administrative tasks such as note-taking or scheduling but draws a line at independent clinical decision-making.
Industry groups, including the Computer & Communications Industry Association and telehealth organizations, had urged the governor to veto the bill, citing concerns that it could be overly broad and limit beneficial AI tools for wellness and support.
Details on the governor’s action remained unclear as of Tuesday. The legislation also adds representatives from professional regulation and medical practice boards to the state’s Artificial Intelligence Advisory Council, which is tasked with providing further recommendations on AI in mental health by January 2027.
Vermont’s move positions it among states seeking to set guardrails on AI applications in health care. Similar efforts have been underway elsewhere, though specifics vary by jurisdiction.
The new provisions take effect according to the bill’s terms once signed, with enforcement primarily through the attorney general under consumer protection authorities. Officials have not released estimates on the number of AI therapy services potentially affected in the state.
