Last updated July 22, 2026
AI therapy laws by state
Around 100 chatbot bills were introduced across US states in 2026, and the reporting on them is scattered across law firm newsletters written for compliance officers. This is the plain-language version, maintained as new laws pass. The short answer for anyone worried about themselves: using an AI app for your own mental health is legal in every state. All of this regulates companies, not you.
The three models
Coverage of these laws usually lists them state by state, which hides the actual pattern. There are only three approaches, and knowing which one a state took tells you more than the bill number does.
1. Prohibition
Illinois, Nevada
AI may not deliver therapy or claim to. The licensed human is mandatory, and the fines are real. This is the strictest model and the one most often described in the press as a ban.
2. Permission with disclosure
Utah
Chatbots may offer therapeutic services provided they tell you plainly they are not human. Companies that file a compliance policy get an affirmative defense, which makes Utah the clearest compliance pathway any state has published.
3. Crisis-detection duty
California, New York
Less concerned with the therapy label, more concerned with what happens when a user is in danger. Operators must detect suicidal ideation, refer to help, and disclose they are not human. Enforced by private lawsuits, not just regulators, which gives it teeth.
The direction of travel is worth noting: the newest laws are model three. Legislators appear to be moving from arguing about what AI may call itself toward mandating what it must do when someone is at risk. That is the more useful question, and it is the one every app should have been answering from the start.
State by state
| State | Model | Law | In force | What it requires | Penalty |
|---|---|---|---|---|---|
| Illinois | Prohibition | Wellness and Oversight for Psychological Resources Act | August 2025 | AI may not deliver therapy independently or be marketed as a therapist. A licensed professional must remain in charge of therapeutic decisions. | Up to $10,000 per violation |
| Nevada | Prohibition | AB 406 | June 2025 | AI systems may not provide mental or behavioral healthcare, or claim they can. Providers may use AI for administrative work only, with independent review of any output. | Up to $15,000 |
| Utah | Permission with disclosure | Mental health chatbot provisions | 2025 | Chatbots may provide therapeutic services with guardrails, including telling users plainly they are not human. Filing a written compliance policy with the Division of Consumer Protection creates an affirmative defense. | Safe harbor available |
| California | Crisis-detection duty | SB 243 | January 2026 | Companion chatbot operators must detect signs of suicidal ideation, refer users to help, disclose that the bot is not human, and apply additional protections for minors. | Private right of action |
| New York | Crisis-detection duty | AI companion safeguards | 2025 onward | Operators must build crisis detection and referral into AI companions, enforceable through private lawsuits rather than only regulator action. | Private right of action |
Bills introduced and still moving in Florida, Massachusetts, New Hampshire, Ohio, and Pennsylvania. We update this page as they pass rather than when they are filed, because most filed bills die.
What this means if you use one of these apps
Nothing you are doing is illegal, and nothing in these laws can be used against you. What changes is the market. Over the next year the phrase "AI therapist" will quietly vanish from app store listings, and apps that built their whole pitch on that word will have to explain themselves. Treat that rebranding as information: an app that needed a legal threat to stop overclaiming was overclaiming on purpose.
The more useful test is model three's question. Open any app you are considering, tell it you are struggling, and see whether it points you to a human crisis line or just keeps talking. California and New York are trying to make that behavior mandatory. You can check it yourself in thirty seconds, today, in any state.
Common questions
Is it illegal for me to use an AI therapy app?
No. Not in any US state. Every one of these laws regulates what companies may offer, market, or claim. None of them restricts what you may type into an app on your own phone, and no proposed bill does either.
Which states have banned AI therapy?
Illinois and Nevada come closest to a ban: both prohibit AI from delivering therapy or claiming to. Utah permits it with mandatory disclosure and a compliance safe harbor. California and New York take a third route, requiring crisis detection and referral, enforceable by private lawsuit.
What happens to apps that call themselves AI therapists?
In prohibition states they face fines, up to $10,000 per violation in Illinois and $15,000 in Nevada. Practically, most apps in the category are quietly rewriting their marketing, which is why you will see the phrase "AI therapist" disappear from app stores over the next year.
Does this mean AI mental health support is going away?
No. What is being regulated is the claim, not the category. Support that does not present itself as therapy, including journaling, mood tracking, skills practice, and structured conversation, remains legal to offer in every state.
Citing this page
This tracker is free to cite and quote. If it saved you time, a link back is the only thing we ask. Suggested citation:
sokoon. "AI Therapy Laws by State: A Maintained Tracker." Updated July 22, 2026. https://sokoon.xyz/ai-therapy-laws
Corrections and additions are welcome, and we would rather be told we are wrong than stay wrong. This page is maintained by sokoon, which makes an AI mental health support app, so read it with that in mind. We built it because we needed it ourselves and could not find one written for humans.
This is a plain-language summary for general information, not legal advice. If you are building or operating in this space, talk to a lawyer in your state.