New York Introduces Legislation to Regulate Use of Artificial Intelligence in Mental Health Care
On August 20, 2025, New York State Senator Kristen Gonzalez introduced the Oversight of Technology in Mental Health Care Act (S.8484). The bill reflects growing interest by state legislators across the country to restrict delivery of unsupervised therapy or mental health advice by AI platforms and desire to strengthen provisions of existing professional standards addressing professional responsibilities, patient privacy, and patient consent.
SB8484 establishes the following key restrictions and safeguards to ensure that use of AI in mental and behavioral care remains human-centric and subject to proper oversight:
Prohibition on Autonomous AI in Direct Care
The legislation expressly prohibits the use of autonomous AI systems that independently provide therapeutic advice, diagnostic guidance, or treatment recommendations related to mental or behavioral health. In short, AI systems cannot replace licensed professionals in providing care directly to patients.
Permitted Integration of AI Under Professional Oversight
Licensed mental health professionals—including therapists, counselors, and other psychotherapy providers—may integrate AI into their practice, but only when:
- The professional reviews and approves all AI-generated recommendations or treatment plans
- Clients are notified about the purpose and role of AI in treatment
- Explicit, informed consent is obtained prior to use of AI. Importantly, consent cannot be secured through broad “terms of use” or passive click-through agreements
- Practitioners must also ensure that all AI-related records are maintained in compliance with state recordkeeping and confidentiality requirements.
Promotion of AI to address administrative inefficiencies and client support
While putting guardrails in place that regulate how a licensed provider uses AI, the bill explicitly encourages use of AI systems to ease administrative burdens and lower administrative costs. Such uses hold the promise of reducing provider burnout and enhancing the client experience.
Allowed uses include:
- Scheduling, billing, and claims processing
- Preparation of patient notes or records
- Aggregation of external resources and referral lists
- Analysis of anonymized data to monitor patient progress or identify trends
Exemptions
The bill exempts religious counseling and peer support services, provided that such services are non-clinical and remain outside the scope of regulated mental health practice.
Perspective and Analysis
Earlier this year, New York passed legislation that prohibits the use of an “AI companion” unless the system is trained to detect suicidal ideation or expressions of self-harm expressed by the user, is capable of referring users to crisis service providers, and provides regular notification that companion is not human. SB 8484 would add an additional level of regulations specifically addressing mental health chatbots, paralleling recently enacted legislation in Utah, Nevada, Illinois.
More than 250 bills addressing use of AI in health care were introduced in state legislatures in 2025. Some of these bills, such as AB489 in California, which prohibits an AI platform from representing itself as a licensed health care professional and empowers the medical and osteopathic boards with enforcement responsibilities, may still become law by year’s end.
These legislative efforts to provide guardrails for AI tools also come as lawsuits are being filed against technology companies and platforms providing mental health services directly to individuals. Last month, the parents of 16-year-old who died by suicide after seeking care through ChatGPT filed a lawsuit against OpenAI in California state court alleging that the company is liable for wrongful death and engaged in deceptive and unfair trade practices.
Technology companies and licensed health care professionals should closely monitor legal trends as state legislatures prepare for 2026 sessions, and companies looking to scale health care AI in highly regulated spaces such as health care must prepare for additional compliance efforts addressing professional licensure, scope of practice, patient consent, explainability, and patient privacy. Health care trade associations and professional societies can play a major role in educating their members to understand how existing health care rules apply to generative AI and should look to educate members on responsible application of AI in clinical, administrative, and research settings, so members stay compliant in this rapidly evolving space.
Professional
For more information, please contact Eric Fish, Stephen Phillips, Andrea Frey, Monica Massaro, or your regular Hooper, Lundy & Bookman contact.
