August 11, 2026  – As artificial intelligence becomes increasingly common in behavioral health, states are moving to establish clearer boundaries around how the technology can be used in mental health treatment.

Recent legislative efforts in Vermont and California illustrate a growing national trend toward requiring human oversight whenever AI is involved in providing mental health services. While supporters believe AI can help improve access to care, lawmakers and clinicians are expressing concern that chatbots should not replace licensed professionals or make independent therapeutic decisions.

In June, Vermont enacted House Bill 816, a law designed to ensure that mental health services are delivered by licensed professionals rather than artificial intelligence alone. The legislation prohibits companies from offering AI-based mental health services unless those services are provided by or under the supervision of a licensed mental health professional. The law also requires clinicians to review and approve any mental health guidance generated by AI before it is delivered as treatment.

The legislation reflects a broader effort to balance innovation with patient safety. AI tools may still be used to support clinicians, but they cannot independently diagnose, counsel, or treat patients.

California lawmakers are pursuing similar goals through Senate Bill 903, which would place additional restrictions on AI chatbots marketed for mental health treatment. The proposal would prohibit companies from advertising AI chatbots as therapists or licensed mental health providers and would require clear disclosures when users are interacting with artificial intelligence rather than a human clinician. The bill also seeks to prevent AI systems from independently diagnosing mental health conditions or making treatment decisions.

Supporters of the California legislation say the measure is intended to protect consumers while still allowing healthcare organizations to use AI as an administrative or clinical support tool. Rather than banning AI outright, the proposal focuses on ensuring that licensed professionals remain responsible for diagnosis, treatment planning, and therapeutic decision-making.

The legislative activity comes as AI-powered chatbots are increasingly being used for emotional support, coaching, and mental health conversations. While some early research suggests purpose-built AI tools may help expand access to behavioral health services when used alongside clinicians, experts continue to warn that general-purpose chatbots can generate inaccurate, misleading, or potentially harmful responses, particularly for individuals experiencing severe mental illness or a mental health crisis.

The debate is unfolding as healthcare organizations continue exploring AI to address workforce shortages and rising demand for behavioral health services. Many providers view AI as a tool that can assist with documentation, patient engagement, and administrative tasks, but they also recognize that treatment decisions require professional judgment and accountability.

Taken together, the actions in Vermont and California suggest states are moving toward a common regulatory framework that allows AI to support behavioral healthcare while preserving the central role of licensed clinicians. As more states consider similar legislation, policymakers are likely to continue debating where technology can safely enhance mental health care—and where human judgment must remain indispensable.