Is AI Therapy Legal? What the New State Laws Actually Ban
Short answer: using an AI app for your own mental health is legal everywhere in the US, and no law on the books restricts you, the user, at all. What is newly illegal in a growing list of states is companies offering or marketing AI as therapy. That distinction matters, it is reshaping the whole category, and it is worth five minutes to understand, because the laws are quietly on your side.
What actually changed in 2025
Illinois moved first. Its Wellness and Oversight for Psychological Resources Act, signed in August 2025, prohibits offering therapy or psychotherapy services through AI, with fines up to $10,000 per violation. A licensed clinician in Illinois cannot delegate therapeutic decisions to an AI, and a company cannot sell AI as therapy at all.
Nevada passed its own restriction the same summer, barring AI systems from providing services that would constitute the practice of psychology or behavioral health care. Utah took a different route, regulating mental health chatbots with disclosure requirements: users must be told clearly that they are talking to software, and their data gets specific protections. More states have bills moving, and federal regulators have started circling the question of generative AI mental health tools.
What the laws target, and what they leave alone
Every one of these laws points at the supply side: what companies may offer, claim, and market. None of them restricts what you may type into an app on your own phone. You cannot be fined for talking to a chatbot about your anxiety in Illinois, and no proposed law goes anywhere near that.
The legal line follows the professional one. Therapy is a licensed activity: diagnosis, treatment planning, clinical accountability. The laws say software cannot perform or impersonate that activity. Support that does not claim to be therapy, journaling, mood tracking, structured conversation, skill practice, remains legal to offer in every state, which is why the honest corner of this category keeps existing while the label inflaters scramble to rebrand.
Why this is good news for you
These laws are doing, with fines, the exact filtering we have been telling readers to do by instinct: any app calling itself an AI therapist has told you something about how it handles the truth. Now the claim is not just a tell, it is a liability. Watch the category rebrand over the next year and you will be able to tell which companies treated the old wording as marketing spice and which ones built their product on it.
sokoon never called itself therapy, and not because we saw the laws coming. The reason is on every page of this site: a therapist is a licensed human, software is not, and a product for your mental health should not start the relationship with an inflated title. The laws just made honesty compulsory. We would like to think we were early.
What to actually do with this
We maintain a state-by-state tracker of these laws, updated as new ones pass, if you want the current picture rather than this snapshot. Nothing changes about whether you can use AI support, tonight or ever. If an app you are considering still markets itself as an AI therapist in 2026, treat that as the red flag the legislatures now agree it is. If you need actual therapy, the licensed human kind, here are the affordable routes to it. And if what you need is the layer around care, support that knows exactly what it is, the honest end of the category is ranked here, our own app included, weaknesses listed.
Legal to use. Free to try. Honest about the rest.
sokoon is AI support that never claimed to be therapy. One free session every day, no card, encrypted by default.