Maine Sends AI Therapy Ban to Governor as Missouri Follows

Two U.S. states moved this week to formally restrict the clinical use of AI in mental health therapy, making it clearer than ever that state legislatures — not Congress — are going to write the first real rules of the road for AI in regulated professions. Maine sent LD 2082 to the governor on April 10, and Missouri's HB 2372 has passed the full House and is sitting with the Senate Committee on Families, Seniors and Health.

Both bills share the same core idea: AI can support licensed mental health providers with administrative work, but it cannot do clinical therapy on its own. Violators face civil penalties of up to $10,000 per incident.

What Maine's LD 2082 Actually Does

Maine's bill prohibits any person from providing, advertising, or offering therapy or psychotherapy services — including through AI — unless the services are provided by a licensed professional. The prohibition extends to marketing: you can't promote an AI chatbot as a therapist, even if you include fine print.

The bill is not a blanket AI ban in mental health care. It specifically carves out administrative support, defined to include managing appointments, processing billing, and drafting logistical communications. With client consent, providers can also use AI to maintain client records and analyze notes from therapy sessions.

The penalty is civil, up to $10,000 per violation, with the amount scaled to the degree of harm and the circumstances. For consumer AI companies offering unregulated "therapy" chatbots to Maine residents, that math gets expensive quickly.

Why This Happened Now

The legislative push didn't come from nowhere. Over the past year, high-profile incidents involving companion chatbots and minors — and a steady drumbeat of research showing general-purpose LLMs giving unsafe advice to users in crisis — pushed mental health regulation to the top of state AI policy agendas.

Maine's approach is deliberately narrow. Rather than trying to regulate AI generally, the state went after a specific failure mode: AI marketed as a substitute for licensed clinical care. That framing made the bill easier to pass than broader AI legislation that often gets stuck in fights over preemption and tech industry pushback.

The American Psychological Association and similar organizations in other states have been lobbying for exactly this kind of targeted legislation. Expect the Maine language to be picked up and adapted by other state legislatures over the next six to twelve months.

Missouri's Path — and Its Complication

Missouri's HB 2372 passed the full Missouri House on April 2 with similar contours: AI banned from therapy, psychotherapy, and mental health diagnosis, with $10,000 per first-violation penalties. A separate companion bill, HB 2368 (and SB 1444 in the Senate), prohibits advertising or representing that AI can act as a mental health professional or provide therapy services.

But Missouri hit a wall that Maine didn't. According to the Missouri Independent, efforts to enact broader AI regulations stalled in the Missouri Senate last week amid fears the legislation could jeopardize nearly $900 million in remaining federal broadband funds for rural internet expansion. Missouri lawmakers are specifically worried that state-level AI rules could trigger clawbacks under federal preemption provisions tied to the infrastructure funding.

This is the dynamic to watch. States that depend heavily on federal tech infrastructure dollars will face pressure to soften AI regulation, even for narrow clinical use cases. The federal government has not passed AI legislation of its own, but it has quietly begun using funding conditions as a tool for discouraging states from getting out ahead.

The Broader State Legislation Wave

The week's activity in Maine and Missouri is part of a much larger pattern. Nebraska's unicameral legislature passed a chatbot bill last week. Maryland's legislature passed a pricing bill regulating AI-set consumer prices. Across the country, states are advancing bills on AI in hiring, insurance, education, and algorithmic decision-making.

State attorneys general are also getting more active. Enforcement actions under existing consumer protection statutes are ramping up against companies making unsupported AI claims, particularly in health and wellness categories.

For companies building AI products that touch healthcare, mental health, or any licensed profession, the compliance environment is no longer theoretical. A multi-state consumer product now needs to track dozens of state-level rules that don't line up with each other — and that conflict is only going to get worse.

What It Means for AI Mental Health Startups

A number of venture-backed companies explicitly market AI therapy or AI companion products with mental health framing. Under Maine's new law, any of them serving Maine residents would need to materially change their marketing, add licensed human clinicians to the loop, or pull out of the state.

The legal gray area that has protected these companies — "we're not therapists, we're wellness products" — is narrowing. Maine's definition targets anyone who "offers" therapy, which courts will interpret more broadly than the startups' current marketing copy anticipates.

Larger consumer AI platforms (OpenAI, Anthropic, Google) have been more cautious about mental health claims, and they've been rolling out safety-focused disclaimers and crisis-detection features for months. They're less exposed legally but still face pressure from state regulators watching how general-purpose models respond to mental health queries.

What's Next

Maine's governor is expected to sign LD 2082 in the coming days; the bill has bipartisan support and no organized opposition. In Missouri, HB 2372 now faces a Senate committee that's under crossfire between mental health advocates pushing for passage and broadband advocates warning about federal fund loss. Several other states — Illinois, New York, California, and New Jersey — have similar bills in various stages.

For enterprise AI buyers, the near-term effect is limited: these laws target consumer products, not back-office AI. For consumer AI developers, the effect is immediate: if your product touches therapy, counseling, or mental health diagnosis, you need to review state-by-state exposure now rather than waiting for a federal framework that isn't coming in 2026.

The bottom line: states are writing the rules for AI in regulated professions faster than Congress can, and the rules are not consistent. Maine just drew a clear line around clinical mental health care. Missouri is trying to draw the same line but getting tangled in federal funding politics. The rest of the country is watching which model works.

Sources: Troutman Privacy Law Update | Crypto News on Therapy Chatbot Bans | Missouri Independent on Broadband Funds