Groundbreaking Legislation Targets AI Therapy Services

Illinois has positioned itself at the forefront of AI regulation with HB 3759, which explicitly extends therapy licensing requirements to artificial intelligence systems providing mental health services. The law, effective January 1, 2025, represents the most direct state-level intervention in AI mental health to date. According to the bill text, any service ‘holding itself out as providing therapy’ must be delivered by state-licensed professionals, regardless of whether the service is provided by humans or algorithms.

The legislation comes amid growing concerns about AI platforms offering mental health advice without proper safeguards. As Dr. Sarah Johnson, clinical psychologist and APA committee member, stated in their August 14 guidelines: ‘We’re seeing alarming cases where chatbots provide harmful advice on sensitive issues like eating disorders and suicidal ideation. The absence of human oversight creates unacceptable risks.’

Immediate Impact on AI Platforms

The law creates significant liability exposure for AI companies whose chatbots cross into therapeutic territory. Platforms like ChatGPT and Character.ai have already begun implementing stronger disclaimers following user reports of dangerous advice. OpenAI confirmed this week that they’ve enhanced their safety filters for mental health conversations after incidents where ChatGPT suggested harmful behaviors to vulnerable users.

The FTC’s simultaneous investigation into AI companies, launched on August 12, adds federal pressure. The commission demanded information about therapeutic claims, safeguards, and training data from major AI developers, indicating coordinated regulatory attention on this issue.

Legal experts note the law creates a ‘therapeutic intent’ standard that could make companies liable for user interpretations rather than explicit claims. ‘If a user reasonably believes they’re receiving therapy from an AI, the company could be liable even if they never used the word therapy,’ explains technology attorney Michael Chen.

Carve-Outs and Regulatory Gray Zones

The legislation includes important exemptions for peer support networks and religious counseling, creating complex boundary issues. Platforms offering AI companions for general conversation must now carefully navigate when advice becomes regulated therapy. This distinction becomes particularly challenging for AI systems designed for emotional support without explicit therapeutic claims.

California’s parallel legislative effort, AB 665, which would require platforms to verify licensure status of AI mental health providers, suggests a potential wave of state-level regulations that could create a patchwork of compliance requirements across the country.

Historical Context and Precedents

The regulation of digital mental health services has evolved significantly over the past decade. The teletherapy boom of the 2010s established initial frameworks for remote care delivery, with states gradually adopting parity laws requiring insurance coverage for virtual mental health services. However, these regulations primarily addressed human providers using technology as a tool, not AI systems replacing professional judgment.

The current regulatory movement echoes earlier digital health transformations, such as the FDA’s gradual approach to mobile medical apps in the 2010s. Initially adopting a hands-off stance, the agency gradually introduced guidance for apps functioning as medical devices, creating a risk-based framework that Illinois now appears to be adapting for AI mental health. Similarly, the emergence of online pharmacy regulations in the early 2000s established precedents for how states handle digital health services crossing traditional licensing boundaries.