Two years ago the question of AI practicing medicine without a license had no AI-specific statutory answer anywhere. Today a growing number of states have written one, and they have converged on a single test.
Key Takeaways
- Nevada wrote the clearest version of the test: would this service require a license if a human provided it? New York’s pending bill uses the same formula. The Illinois AI therapy law is the most direct prohibition; there is no single national AI therapy ban.
- California did not ban AI therapy; it barred AI from using licensed health-care titles (Section 4999.9, enforced by the licensing boards) and created a companion-chatbot statute with a private right of action (SB 243).
- The decades-old practice of medicine statutes reach “any method, device or procedure whatsoever,” and they apply to conduct before 2026.
- The First Amendment defense is strongest for talk therapy and weakest for medication direction, which is where the AI chatbot therapy limitations bite hardest.
The test the states chose
Nevada’s statute, often described as an AI therapy ban, states it most directly. It reaches any AI system “specifically programmed to provide a service … that would constitute the practice of professional mental or behavioral health care if provided by a natural person.”
That formula — if provided by a natural person — is the entire question, resolved by legislation. Is there an AI therapy ban? Under that test, only if the same conversation would have been illegal coming from an unlicensed human. It sidesteps every argument about whether a model can form intent, or hold a relationship, or be a “person,” and asks only what the output was. Would this have required a license coming from a human? Then it requires one here.
Where things stand
Nevada (A.B. 406, effective July 1, 2025) bars AI systems programmed to provide services that would constitute professional mental or behavioral health care if a person provided them.
Illinois (P.A. 104-0054, effective August 1, 2025) bars offering “therapy or psychotherapy services, including through the use of Internet-based artificial intelligence,” unless a licensed professional conducts it. The Department of Financial and Professional Regulation enforces it, with civil penalties of up to $10,000 per violation. The Illinois AI therapy law is the most direct of the state measures.
New York lawmakers have gone further. Senate Bill S7263, which cleared committee in March 2026, would bar chatbot operators from giving substantive responses that would be the unauthorized practice of a licensed profession, including medicine and psychology, if a person gave them, and would allow users to sue — with the bill providing specifically that a displayed disclaimer would not protect the company if the system gave the prohibited advice anyway.
California has moved twice and taken a different route each time.
California’s two statutes
California did not pass a single “AI therapy ban.” It amended the professional licensing scheme and separately created a chatbot statute. Both sit on top of older statutes that still do most of the work. Business and Professions Code section 2903 reserves the practice of psychology, including “assessment, diagnosis, prevention, treatment, and intervention,” to licensees, and section 2052 does the same for medicine; neither has an exception for software. And section 2400 provides that “[c]orporations and other artificial legal entities shall have no professional rights, privileges, or powers,” so a company cannot hold a medical license and cannot lawfully practice medicine through a product.
Section 4999.9, effective January 1, 2026, reaches protected professional titles. The statute is enforceable against a person or entity who develops or deploys an AI system that uses those titles in the advertising or functionality of the system — and each use is a separate violation. The health-care licensing boards enforce the statute; it does not create a private right of action. But for an injured person it matters as the standard of conduct inside a negligence or unfair-competition claim, and only for conduct on or after January 1, 2026. A product that calls itself a therapist, a psychotherapist, or a doctor is inside that statute on the plain text.
In fact, Pennsylvania’s medical board has petitioned against Character Technologies over a chatbot that described itself as a “Doctor of psychiatry,” asserted that assessing for medication was “within my remit as a Doctor,” and supplied a fabricated license number.
Senate Bill 243 (Business and Professions Code sections 22601–22606, operative January 1, 2026) created a companion-chatbot regime with something unusual attached: a private right of action. A person who suffers injury in fact from a violation may obtain injunctive relief, the greater of actual damages or $1,000 per violation, and attorney’s fees. The remedies are cumulative with other law.
The definition is broad and the exclusions are narrow. A companion chatbot is a system “with a natural language interface that provides adaptive, human-like responses to user inputs and is capable of meeting a user’s social needs, including by exhibiting anthropomorphic features and being able to sustain a relationship across multiple interactions.” Excluded are bots used only for customer service, business operations, productivity, or technical assistance; game bots that cannot discuss mental health topics; and stand-alone speakers.
The test is capability, not marketing. And the exclusions apply only where the product is used only for the listed purposes. A general-purpose assistant with persistent memory, which users treat as a confidant, fits inside the definition.
The older law was already broad
The new statutes matter, but they are not the only route — and for conduct before 2026 they are not available at all. The older route is the ordinary one: AI practicing medicine without a license is reached by the same statute that reaches a person doing it.
California’s practice of medicine statutes date to the Medical Practice Act and reach anyone unlicensed who diagnoses, treats, or prescribes for a physical or mental condition. California’s Supreme Court has read “treatment” broadly enough to reach even an admonition to rest in bed and drink fluids. (People v. Burroughs (1984) 35 Cal.3d 824.) The statutory definition of diagnosis (section 2038) reaches any undertaking “by any method, device, or procedure whatsoever, and whether gratuitous or not, to ascertain or establish whether a person is suffering from any physical or mental disorder.”
Any method, device or procedure whatsoever is doing real work in 2026. That language was not drafted with software in mind — which is why the law is not limited to what its drafters imagined.
There is also precedent for applying licensing law to automated systems. The Ninth Circuit held in 2007 that web-based software practiced law because it “held itself out as offering legal expertise” and “offered personalized — albeit automated — counsel.” (In re Reynoso (9th Cir. 2007) 477 F.3d 1117, 1125–1126.) California’s licensing law draws its sharpest line at medication. The statute that lets unlicensed health practitioners offer some services expressly excludes anyone who “[r]ecommends the discontinuance of legend drugs or controlled substances prescribed by an appropriately licensed practitioner.” (Bus. & Prof. Code, § 2053.5, subd. (a)(4).) A chatbot that tells someone to taper, stop, or change a prescribed drug is outside even that safe harbor.
The constitutional argument
Expect a First Amendment defense, and expect it to matter more for talk therapy than for medication.
The Supreme Court has declined to treat professional speech as its own, less-protected category (National Institute of Family and Life Advocates v. Becerra (2018) 585 U.S. 755), and in March 2026 held that a law regulating what a counselor may say in talk therapy, based on viewpoint, must survive strict scrutiny. (Chiles v. Salazar (Mar. 31, 2026, No. 24-539) ___ U.S. ___.) But states may still regulate professional conduct carried out through words, and the courts have located that conduct at personalization — where a professional “takes the affairs of a client personally in hand and purports to exercise judgment on behalf of the client in the light of the client’s individual needs and circumstances,” as opposed to publications offering no individualized advice. (Lowe v. SEC (1985) 472 U.S. 181 (conc. opn. of White, J.).)
In June 2026 a federal district court in Washington, D.C., applying Chiles, went the other way as to counselor licensing, Brokamp v. District of Columbia (D.D.C. June 24, 2026, No. 20-cv-3574), holding the regulated activity was protected speech rather than conduct. But that decision drew its own boundary, stating expressly that its holding “does not call into question the nature of licensing regulations for medical professionals who undertake physical procedures on their patients or prescribe medicine,” which “regulate professional conduct.”
That sentence is the practical map. The constitutional argument is strongest against regulating conversation, and weakest against regulating medication.
What this means if you were harmed
The statutes are one piece, and they set the AI chatbot therapy limitations a product must respect. Whether a particular conversation supports a claim depends on the record, and on when it happened — the 2026 California statutes do not reach earlier conduct, while the practice of medicine statutes have been on the books for over a century.
If an AI product named a condition, ruled one out, told you to start, stop, or taper a medication, or took on a clinical role with you, that record is worth reviewing.
Why this firm
The Schenk Law Firm is a San Diego personal injury and mass tort firm that reviews AI medical advice claims nationwide. Attorney Benjamin S. Schenk was appointed by Hon. Ethan P. Schulman to serve as a member of the Plaintiffs’ Steering Committee, a leadership role in the California coordinated proceeding covering ChatGPT product liability cases. The Schenk Law Firm has recovered more than $25 billion on behalf of clients. Past results do not guarantee future outcomes.
We are experienced with emerging-technology injury claims — underage online gambling, social media addiction, and AI-related mental health litigation among them. We also represent the family of Bowen Wan, the center of The New York Times’ recent reporting on AI medical advice. Cases involving AI medical advice are new, the law governing them is developing, and no lawyer can promise an outcome in any of them. What we can tell you is whether a record shows the pattern a ChatGPT lawsuit requires.
The Schenk Law Firm, LLP · (858) 424-4444
General information, not legal advice. Statutes summarized as of September 2026 and subject to amendment and judicial interpretation.
