State legislators across 11 states have enacted more than a dozen new laws in 2026 to regulate artificial intelligence in healthcare, responding to mounting pressure from patients and medical professionals. The laws, detailed in the Transparency Coalition's 2026 Mid-Year State AI Legislation Report, focus on two main areas: restricting AI in insurance authorization decisions and banning AI chatbot-based therapy services.
The new measures address growing concerns over inappropriate deployment of AI for Healthcare, setting new rules for when and how automated systems can interact with patients.
Limiting AI use in medical authorizations
Seven new state laws require human oversight when health insurers use AI to approve or deny procedures. Most mandate that a licensed physician or healthcare professional make the final determination, and prohibit decisions based solely on group data rather than the patient's individual medical history and circumstances. AI for Insurance authorization decisions must now include a human in the loop.
- Alabama (SB 63): Insurers must disclose AI use, and denials must be made by a licensed professional. Enacted April 17, effective Oct. 1, 2026.
- Colorado (HB 1139): Denials must be reviewed by a licensed clinician or physician, and decisions cannot rely solely on group data. Enacted June 2, effective Jan. 1, 2027.
- Georgia (SB 444): Prohibits basing coverage decisions solely on AI systems or software tools. Enacted May 5, effective Jan. 1, 2027.
- Illinois (SB 3114): Bans payors from using automated processes to downcode a claim without evaluating all provider-submitted information. All downcoding determinations must be made or reviewed by a natural person. Awaiting governor's signature.
- Iowa (HF 2635): Insurers may use AI for initial review but cannot deny, delay, or downgrade a request solely on AI. Enacted May 13.
- Utah (SB 319): Requires independent medical judgment when reviewing adverse preauthorization determinations, disclosure of AI use, and time limits on authorization decisions. Enacted March 19, effective Jan. 1, 2027.
- Washington (SB 5395): Only a licensed physician or health professional may deny a prior authorization request. AI shall not be the sole means to deny, delay, or modify care. Enacted March 23, effective June 11.
Therapy chatbot bans
Five states now prohibit offering mental health therapy through AI chatbots. The bans build on earlier 2025 laws in Illinois and Nevada, with mental health professionals playing a central role in pushing for the restrictions. Most new laws tie the prohibition to state licensing authority, making it illegal for a person or business to advertise or provide AI-based psychotherapy unless a licensed professional is involved.
- Colorado (HB 1195): Bans AI therapy chatbots and restricts AI use by licensed psychologists, counselors, and social workers. Enacted June 3, effective Aug. 12.
- Maine (LD 2082): Offering AI therapy without a licensed professional is an unfair trade practice. Enacted April 13, effective July 29.
- Rhode Island (H 7349 / S 2197): Prohibits advertising or providing therapy unless conducted by a licensed provider; licensed providers cannot use AI to make independent therapeutic decisions. Enacted June 22, effective Jan. 1, 2027.
- Tennessee (SB 1580): Prohibits developing or deploying an AI system that advertises itself as a qualified mental health professional. Enacted April 1, effective July 1.
- Vermont (H 816): Protects individuals from psychological harm by ensuring therapy is delivered by professionals, not independently by AI. Enacted and effective June 17.
Patient consent required
Iowa enacted HB 475, requiring licensed healthcare professionals to verbally tell a patient before using any recording device, software, or service to record and transcribe a clinical appointment with AI. Violations are handled by professional licensing boards and do not trigger civil liability absent gross negligence or willful misconduct. The law takes effect Aug. 1.
Scope of practice limitations
Utah passed SB 150, which specifies that AI does not qualify as an innovation or technology upgrade within a medical clinic's scope of practice. The definition of "innovation" explicitly excludes any technology upgrade that solely provides advice or treatment to a patient without a healthcare practitioner-patient interaction. The law was enacted March 24.
Why this matters for healthcare professionals
These laws create a patchwork of new compliance obligations that vary by state. Clinicians and practice managers must verify that prior authorization denials involve a licensed reviewer, especially in Colorado, Washington, and Utah when those laws take effect in 2027. Providers in Iowa need consent workflows for AI recording of patient visits. Mental health professionals should audit any AI tools to ensure they are not making independent therapeutic decisions, which would run afoul of bans in five states. For multi-state practices, mapping each jurisdiction's effective date and specific human-review requirements is now an operational necessity.
Your membership also unlocks:
