WeVote
Log In

Bill

Bill

H 644

An act relating to regulating the use of artificial intelligence in the provision of mental health services

2025-2026 Regular Session Introduced by Ela Chapin and 12 co-sponsors

Vermont bans using AI to deliver or guide mental health therapy, allows AI only for admin tasks or transcription with explicit consent and professional oversight.

Read first time and referred to the Committee on Health Care
0
WeVote Research Nonpartisan
Bill Summary · H 644

Summary of H.644 (2025-2026) — Vermont

Purpose and intent

  • Aim: Prohibit the use of artificial intelligence (AI) in delivering mental health services, with narrow exceptions, and regulate AI use by mental health professionals.
  • Overall goal: Safeguard Vermonters seeking mental health care from potential psychological harm, including risk of suicide, by ensuring services are delivered by qualified mental health professionals rather than AI systems.

Key provisions and changes

Prohibition on AI-delivered mental health services

  • It is unlawful for a person, corporation, or other entity to offer, provide, or advertise mental health services in Vermont that use AI in whole or in part, except as authorized by specific provisions.
  • The act defines AI broadly to cover machine-based systems, software, or algorithms capable of perceiving, processing, and acting on data to imitate intelligent behavior (including NLP, pattern recognition, predictive analytics, etc.).

Definitions relevant to scope

  • Mental health services: Includes peer support, counseling, therapy, psychotherapy, diagnosis, treatment, and ongoing recovery support.
  • Therapeutic communication: The direct clinical interaction aimed at diagnosing or treating mental health conditions or guiding recovery.
  • Administrative support: Non-therapeutic tasks assisted by AI (e.g., scheduling, billing, notes, data analysis for progress tracking).

Prohibited uses and permitted uses (in professional practice)

  • Prohibited: Mental health professionals may not use AI to make therapeutic decisions, issue direct therapeutic communications, generate treatment plans or recommendations, or detect/interpret emotions or mental states.
  • Permitted (under strict conditions):
    • Use of AI for administrative support is allowed, provided the mental health professional reviews and retains responsibility for all outputs, data use, and tasks performed by the AI.
    • Use of AI for transcription/recording is allowed only if:
    • The patient/client (or their guardian) is informed in writing of the purpose and that transcriptions are subject to confidentiality rules.
    • Consent is obtained from the patient/client or guardian.

Confidentiality and consent

  • Transcriptions or recordings produced by AI remain subject to confidentiality and disclosure prohibitions under Vermont law (18 V.S.A. §§ 1881 and 7103).

Professional regulation and enforcement

  • Unprofessional conduct: Misuse of AI by a licensed mental health professional constitutes unprofessional conduct under 3 V.S.A. § 129a (with scope to include actions both inside and outside the state).
  • Consumer protection enforcement: Violations of the AI provisions are treated as violations of Vermont’s Consumer Protection Act (9 V.S.A. ch. 63). The Attorney General can enforce, and private parties can sue, with potential civil penalties.

Civil penalties

  • Each violation carries a civil penalty of $10,000 (as referenced in 9 V.S.A. § 2461).

AI regulatory framework for licensed professions

  • Addition of new Chapter 120 (Artificial Intelligence in Regulated Professions) to 26 V.S.A.
  • Subchapter 2 focuses on Use of AI by Mental Health Professionals:
    • Defines administrative support, consent, mental health professionals (broadly including physicians, APRNs, psychologists, social workers, counselors, peer specialists, and other licensed/non-licensed practitioners who provide mental health services).
    • Reiterates that AI cannot be used to make therapeutic decisions, generate direct therapeutic communications, or create treatment plans in therapeutic settings.
    • Permits limited AI use for administrative tasks and transcription/recording with explicit informed consent and patient notification.
    • Requires professionals to remain responsible for outputs and data.

Who is affected

  • Mental health professionals and regulated/rostered providers in Vermont who deliver mental health services.
  • Entities offering or advertising mental health services in Vermont that could utilize AI.
  • Patients/clients receiving mental health services, who would gain stronger protections around consent, confidentiality, and awareness of AI use.

Effective date and timeline

  • The act provides that it shall take effect on passage (immediate to near-immediate effect if enacted and signed).

Key implications

  • Strong precautionary framework against AI-based therapy delivery.
  • Clear delineation between permissible AI administrative tools (with oversight) and prohibited therapeutic AI applications.
  • Enhanced consumer protection with potential penalties for violations.
  • Expanded regulatory architecture for AI in regulated mental health professions, including definitions and consent requirements.

Compiled from official sources — confirm details with the bill’s official record.

Sign in to ask a question.