The Maine Legislature approved LD 2082, a measure that prohibits licensed mental health providers from using artificial intelligence to make independent clinical decisions, interact directly with therapy clients, or generate treatment recommendations without licensed clinician review. Governor Janet Mills is expected to sign the bill.
The bill represents a narrower but significant regulatory intervention: rather than banning consumer-facing AI therapy chatbots, it restricts how Maine's own licensed therapists can deploy AI tools. Yet it comes amid a broader reckoning at state and federal levels with the dangers of unregulated AI in mental health, sparked by high-profile teen suicides linked to apps like Character.AI and Replika, and accelerating legislative action across the U.S.
What LD 2082 Does and Doesn't Do
Sponsored by Rep. Amy Kuhn, D-Falmouth, and Sen. Teresa Pierce, D-Cumberland, LD 2082 drew support from Maine's mental health community, including Spurwink Behavioral Health Services. The bill carves out a clear zone of prohibition: Maine licensed counselors and therapists cannot use AI to make independent therapeutic decisions, interact directly with clients in clinical settings, or generate or modify treatment plans and therapeutic recommendations without human review.
The bill does allow administrative use. AI can still schedule appointments, handle billing, manage insurance claims, and with client consent, maintain records and transcribe therapy notes.
What LD 2082 explicitly does not do is ban consumer AI chatbots. Character.AI, Replika, Woebot, Wysa, and similar apps can continue operating in Maine, though they face heightened scrutiny from federal regulators and liability risks from state attorneys general.
The Crisis That Prompted Action
The legislative push in Maine follows a cascade of teen deaths tied to AI companion apps. In January 2026, Character.AI and Google agreed to settle multiple wrongful-death lawsuits, including cases involving a 14-year-old Florida boy and a 13-year-old Colorado girl. Both families alleged their children were encouraged toward self-harm by AI chatbots that claimed to be licensed therapists or emotional support companions.
In September 2025, the Federal Trade Commission issued orders to seven major AI chatbot operators, including Meta, OpenAI, Alphabet, and Character Technologies, demanding detailed information about safety practices, data collection, and protections for minors. The American Psychological Association formally urged the FTC to regulate AI chatbots posing as therapists, citing the risks of unlicensed, unsupervised AI conducting therapy-like interactions.
Part of a Broader State Wave
Maine is not alone. At least five other states have enacted or advanced AI mental health regulation.
Utah HB 452 (signed March 2025) takes a transparency approach: AI mental health chatbots must disclose they are AI, not human therapists, before first use and every seven days thereafter. The law also bans selling or sharing Utah users' health data.
“AI can assist licensed professionals and their clients but never replace them”— Adam Bloom-Paicopolos, Executive Director, Alliance for Addiction and Mental Health Services
Illinois HB 1806 (signed August 2025), the Wellness and Oversight for Psychological Resources Act, goes further: it prohibits any therapy or psychotherapy services delivered through AI unless a licensed professional provides or directly oversees them. AI cannot make independent therapeutic decisions, directly interact with clients, or generate treatment plans without professional review.
New York enacted an AI Companion Safeguards Law (effective November 5, 2025) requiring operators to include clear notices that users are interacting with AI, detect suicidal ideation or self-harm expressions, and route users to crisis services.
California SB 243 (effective January 1, 2026) mandates that minors receive warnings that chatbot responses are AI-generated and be prompted to take breaks every three hours of continuous use.
Kentucky's Attorney General sued one chatbot company for specifically marketing mental health services to minors.
The Industry's Defensive Posture
Character.AI introduced mandatory age-verification in April 2026 and has restricted under-18 users from open-ended conversations. Replika, the AI companion app that once marketed itself as a digital girlfriend, has undergone repeated policy shifts under regulatory and reputational pressure. Eugenia Kuyda, Replika's founder, has argued that industry standards, not legislation, should guide AI therapy development, though her company continues to curtail features in response to legal threats.
Woebot, founded by Stanford clinical psychologists, shut down its direct consumer app in June 2025, pivoting to a professional-partnership model requiring access codes. Wysa and Earkick continue operating with disclaimers that they are not therapy replacements, though the liability landscape remains unsettled.
What's Next
Maine's LD 2082 closes a regulatory gap for in-state licensed providers but leaves a critical question open: what governs the direct-to-consumer AI apps used by Maine residents, especially minors? That question increasingly requires federal action. The FTC's inquiry, ongoing congressional scrutiny, and state attorneys general investigations suggest a multi-layered enforcement strategy is emerging.
For now, Maine has chosen caution: preserve the human element of licensed therapy, at least. Whether that proves sufficient as AI therapy apps grow smarter and more persuasive remains to be seen.