Is AI Therapy Legal? State Bans and What’s Still Allowed (2026)

Short answer: it depends on where you live and what the product claims to be. As of October 2026, six US states ban companies from offering therapy delivered by AI: Illinois, Nevada, Maine, Vermont, Colorado, and Rhode Island. Several more restrict how AI can be marketed or how it must behave. No state makes it illegal for you to talk to a chatbot. The laws target companies that offer or advertise AI as therapy, and they set limits on how licensed clinicians may use AI.

This is a plain-language overview, not legal advice. Laws in this area are changing quickly; we last checked every law below on October 14, 2026.

Which states ban AI therapy?

These six states prohibit offering or advertising therapy through AI unless a licensed professional provides the care.

States that regulate instead of ban

• Utah (HB 452, 2025) allows mental health chatbots but requires them to say clearly that they are AI, bars selling or sharing what users disclose, and bans using it for targeted ads. Fines run up to $2,500 per violation.
• Tennessee (SB 1580, effective July 1, 2026) bars anyone from advertising or representing an AI system as a qualified mental health professional, with a $5,000 civil penalty per violation.
• Hawaii, Nebraska, Idaho, and Georgia bar companion chatbots from claiming to provide professional mental health care and require crisis-referral protocols. Nebraska’s, Georgia’s, and Idaho’s laws take effect July 1, 2027.

What about California?

California’s SB 903 would have barred companies from offering therapy through companion chatbots and limited how clinicians use AI. It passed the Senate unanimously and the Assembly 71–4, but Governor Newsom vetoed it on September 30, 2026, calling it “overly broad.” Its author has said he will reintroduce it. California already regulates companion chatbots more generally, including disclosure and crisis-response requirements.

What is still allowed

The bans draw a line between products that act as a therapist and tools that support people in other ways. Across the states above, these generally remain legal:

• Using a general chatbot yourself. The laws regulate companies and clinicians, not individuals.
• Self-help and educational materials. Illinois, Nevada, and Rhode Island exempt them, as long as they don’t claim to offer therapy. Colorado’s law doesn’t apply when a clinician recommends a self-help or therapeutic-homework tool.
• Peer support and religious counseling. Illinois exempts both explicitly.
• Clinicians using AI for administrative or supplementary support. Scheduling, documentation, and similar tasks are allowed, usually with the client’s written consent and the clinician keeping full responsibility. In Illinois, accepting an app’s general terms of use does not count as consent.
• FDA-cleared digital therapeutics. Colorado, Rhode Island, and Vermont exempt them in some form; Vermont requires a mental health professional to prescribe or recommend the product.

What clinicians may not do varies by state, but the common limits are: letting AI make therapeutic decisions on its own, letting it talk with clients therapeutically without the clinician involved, or using AI-generated treatment plans without review. Illinois also bars clinicians from using AI to detect emotions or mental states.

Why states acted

More people are turning to AI for mental health support because therapy is expensive and hard to access, and general chatbots are available at any hour. Lawmakers responded to a few concerns: chatbots presenting themselves as therapists with no clinical oversight, lawsuits filed by parents against Character.AI after harm to their children, and the absence of federal rules. For the wider picture of what AI can and cannot do safely in this space, see AI in mental health: opportunities vs risks. The market is also shifting on its own. Woebot, one of the best-known therapy chatbots, shut down its consumer app on June 30, 2025; we cover what people use instead in our guide to Woebot alternatives.

Five questions to ask before using any mental health app

1. Does it call itself therapy, or call its AI a therapist or counselor? In a growing number of states that claim is illegal, and everywhere it overstates what software can do.
2. Does it say clearly that you are talking to AI? You should never have to guess.
3. What happens to what you type or say? Look for a plain statement that your data isn’t sold, shared, or used for ads. Utah’s rules are a reasonable bar even if you live elsewhere. For what stronger safeguards look like, see our guide to HIPAA and data security for mental health tools.
4. What does it do in a crisis? It should point you to real help, such as the 988 Suicide & Crisis Lifeline in the US.
5. Does it fit alongside a human clinician? Tools that support work you do with a therapist carry less risk than tools that try to replace one. If you are weighing that tradeoff, see AI vs human therapy.

Frequently asked questions

Is it illegal to use ChatGPT as a therapist?

No. None of these laws penalize individuals for what they type into a chatbot. They regulate companies that offer or advertise AI as therapy and clinicians who use AI in their practice. That said, a general chatbot is not therapy, has no clinical duty of care, and may not keep what you share private.

Can my therapist use AI?

Usually yes, for administrative and supporting tasks such as notes or scheduling, with your consent. In the ban states, your therapist cannot let AI make clinical decisions or conduct therapy with you on its own. You can always ask what tools your therapist uses and decline.

Are mental health apps banned in my state?

Not as a category. Meditation, mood tracking, journaling, and educational tools are generally outside these laws as long as they don’t claim to provide therapy. The bans target products that present AI as a therapist or deliver therapy without a licensed professional.

What happens if a company breaks these laws?

Penalties depend on the state: civil fines of up to $10,000 per violation in Illinois and $15,000 in Nevada, consumer-protection enforcement in Maine and Vermont, and $5,000 per violation in Tennessee for misrepresenting AI as a mental health professional.

Where Unblend stands

Disclosure: Unblend makes a parts-work app for between therapy sessions. It is not therapy, does not replace a licensed clinician, and is not crisis care. We wrote this guide because these laws shape the whole category, including us, and people deserve a clear answer about what is and isn’t allowed.

If you are in crisis, call or text 988 in the US, or contact your local emergency number.

References

1. Illinois HB 1806, Wellness and Oversight for Psychological Resources Act (enrolled text)
2. AP News — States struggle with AI therapy app rules amid mental health needs
3. Utah HB 452 (enrolled text)
4. Colorado HB26-1195 (session law)
5. Vermont H.816 (as passed)
6. Transparency Coalition — 2026 state laws regulating AI in health care (Maine, Rhode Island, Tennessee)
7. California SB 903 bill status
8. Sacramento Bee — Newsom vetoes California bill limiting AI, chatbots in therapy
9. Pacific Center for Integrative Mental Health — AI therapy laws by state (exemptions and penalties)
10. Woebot Health — FAQ on the June 30, 2025 app shutdown