The Week Everything Moved: 14 States Advance AI Mental Health Legislation

Three state legislatures passed AI bills last week. Nebraska's unicameral legislature passed LB 525, bundling the Conversational AI Safety Act with an agricultural data privacy law. The chatbot provisions prohibit AI from claiming to provide professional mental or behavioral health care, require disclosure when a user might believe they are speaking with a human, and add safeguards for youth accounts. Violations carry $1,000 civil penalties enforced by the state attorney general. The bill heads to the governor's desk.

Maine passed LD 2082, which bans AI therapy unless delivered by a licensed professional. The bill allows AI for administrative support and, with consent, record-keeping. Governor Janet Mills received the bill on April 10. The legislature adjourns tomorrow. If she does not act, the bill dies with the session.

Maryland passed HB 895 targeting AI-driven surveillance pricing. Not mental health, but the same legislative wave.

Beyond the three that passed: Connecticut's 97-page omnibus SB 5 cleared the Joint Judiciary Committee. Minnesota's HF 3893 passed its second committee. California advanced three healthcare AI bills in a single week. Oklahoma's SB 1521 passed a House committee. Louisiana's HB 475 requires verbal client consent before AI transcription. Missouri's HB 2372 carries a therapy chatbot ban with $10,000 penalties. Hawaii placed two chatbot bills on 48-hour notice. Tennessee's Senate passed SB 837, declaring AI cannot be granted legal personhood.

The Future of Privacy Forum tracks 98 chatbot-specific bills across 34 states. Over 600 AI bills have been introduced in 2026 state sessions. The federal CHATBOT Act remains in committee. States are not waiting.

Researchers at Aalto University in Finland tracked the public Reddit activity of nearly 2,000 Replika users for two years, one year before and one year after they first mentioned using the AI companion. They compared those users against matched control groups using causal inference techniques designed to isolate the effects of the AI from everything else in those people's lives. Then they interviewed 18 active users.

What they found was a paradox. Users' posts increasingly revolved around their AI relationships. But their language showed more signals of loneliness, depression, and suicidal thoughts than the comparison groups. Not less. More.

The AI companion had no friction. No disappointment. No demand. And that frictionlessness became the trap. When a human relationship requires effort and the AI never does, the human relationship starts to feel like the problem. Over time, people stop reaching out.

This is one of the first causal, longitudinal examinations of AI companion mental health impact at scale, grounded not in surveys but in the unfiltered language of real users. It will be presented at CHI 2026 in Barcelona, the leading international conference on human-computer interaction.

Talkiatry raised $210 million in a Series D in February, the largest behavioral health raise of 2026. Grow Therapy closed $150 million. Salma Health emerged from stealth with $80 million for transcranial magnetic stimulation and treatment-resistant depression. Then the consolidation started: Spring Health acquired Alma, combining $686 million in total funding. PitchBook named Spring Health a potential IPO candidate. NOCD acquired trauma provider Rebound Health and rebranded as Noto.

Somethings raised $19.2 million for youth peer support. Big Health raised $23.7 million for its FDA-cleared digital treatments SleepioRx and DaylightRx, among only nine treatments in CMS's new Digital Mental Health Treatment category with billing codes G0553 and G0554.

The pattern is clear. Investors who got burned in 2021 and 2022 by low-acuity wellness apps have retrained their thesis. They are backing proven unit economics, insurance reimbursement models, and specialty verticals: pediatric, maternal, substance use, interventional psychiatry. When CMS created a reimbursement pathway for FDA-cleared digital treatments, it gave investors a revenue model that depends on the same payer infrastructure that funds every other medical treatment in the country.

Minnesota House File 3893 would prohibit AI psychotherapy unless delivered by a licensed human professional. It has passed two committees and been referred to a third. But the hearing on March 18 at the state Capitol split down a line that matters more than party.

Dr. Steve Girardeau, legislative chair of the Minnesota Psychological Association, supported the bill. "Do we really want to program a computer algorithm to help somebody through their worst moments?" he said. Between him and the opposition sat Eric Mische, CEO of Suicide Awareness Voices of Education: "Psychotherapy is not casual interaction. It requires training, judgment, and accountability."

Ashley Kjos, a licensed psychologist and CEO of Woodland Centers serving rural west-central Minnesota, opposed the bill. Her argument was not that AI therapy is safe. Her argument was that the bill targets the wrong layer. "Regulate the companies that develop AI mental health tools," she said. "Prohibit AI from presenting itself as a licensed clinician. Require transparency." In a region where there are not enough human therapists to serve the communities that need them, banning the only available tool while leaving the shortage untouched creates a different kind of harm.

This is the paradox that runs through every therapy bot ban in the country. The Therapist in the Loop resolves it: not AI alone, not AI banned, but AI supervised by a licensed clinician with escalation protocols and accountability built in. Minnesota has not written that bill yet.

On April 7 and 8, in Nashville, HMP Global hosted the Behavioral Health AI Summit: the first conference in the United States dedicated exclusively to artificial intelligence in behavioral health. Not a panel at a larger event. A summit.

Yesterday, Becker's Healthcare ran its AI Summit for hospital and health system executives. Tomorrow and Tuesday, April 15 and 16, the Becker's Behavioral Health Summit opens at the Hyatt Regency Chicago with over 70 speakers and sessions on AI-transformed referrals, technology in behavioral health from promise to practice, and the next era of care from crisis response to continuity.

Ross Harper, CEO of Limbic, the AI triage platform used by 63 percent of NHS England's therapy program and expanding into 13 US states, put it plainly: "Clinically validated AI agents will be the great unlock. They have the potential to really scale clinician supply and amplify human expertise."

Two years ago, a behavioral health AI summit would not have filled a room. This week, three events in eight days across two cities drew hundreds of executives, clinicians, and investors. A field does not get its own summit circuit until it has arrived.

Casey is a voice-first AI-native mental health EHR with Casey Life and Peer AI Coach supervised by licensed therapists. It completed 1.78 million structured safety test executions with 100% accuracy across all crisis-adjacent categories and zero false negatives on active crisis scenarios as part of pre-FDA safety validation for the 510(k) pathway.

Between sessions, the client has access to a peer coach that operates within the therapist's treatment plan. When risk escalates beyond what the coach can support, the system connects the client to their therapist. Every interaction is logged. Every escalation is documented. The therapist is always in the loop.

▶ Watch the Demo