A pair of Colorado bills to restrict the use of Artificial Intelligence in healthcare is the first AI regulation moving through the statehouse this year. The policies include banning therapists from using chatbots with patients, and placing more restrictions on how generative AI is used in the medical field.

House Bill 1195, focuses on guardrails for how licensed mental health professionals can use AI in psychotherapy and would prohibit psychologists, counselors, and social workers from using AI chatbots to communicate directly with clients. They also could not use AI to generate treatment plans or therapeutic recommendations without approval and review from a regulated mental health professional.

“I think mostly the number one thing it does… is making sure that therapy stays between humans,” said Democratic Rep. Gretchen Rydin, of Littleton, a main sponsor and a practicing licensed therapist.

Rydin said providers could still use AI for administrative and support tasks, such as note-taking or scheduling appointments, but if a therapist uses AI to record or transcribe therapy sessions, it would require written informed consent from the client.

Companies would also be restricted from marketing artificial intelligence tools as psychotherapy, and it would be illegal to provide or advertise those services in Colorado unless they are delivered by a regulated professional.

The measure sailed through committee with unanimous support.

Another piece of legislation, House Bill 1139, would ban health insurance companies from using AI systems exclusively to deny coverage and requires companies to take individual medical histories and circumstances into account when making coverage decisions. It also requires decisions to be reviewed by a qualified human expert.

“The goal of the bill is simple,” said Blair Skinner, president of the Mental Health Advocacy Network of Colorado, who testified in support of it at the hearing. “AI should not be allowed to deny treatment coverage that’s so desperately needed without a trained clinician involved.”

The measure would also require clinicians to let patients know exactly when and how they’re using AI in the course of providing care and prohibit chatbots from presenting themselves as human licensed clinicians.

Democratic Rep. Junie Josephs of Boulder, one of the main sponsors, emphasized that the legislation is not intended to ban AI in health care.

“It sets the rules of the road by requiring transparent disclosure of how it is used and what human oversight exists and prohibits misleading representations,” Josephs said. “Especially when a tool presents itself as equivalent to a licensed professional or implies confidentiality without clear limits.”

House Bill 1139 passed on a party-line vote.

Neither proposal evoked strong opposition, though many players in the tech, medical and behavioral healthcare sectors said they are watching closely and want amendments to the policies, including the Colorado Technology Association.

“CTA members support clear guardrails that protect Coloradans,” president and CEO Brittany Morris Saunders wrote in a text message to CPR News. “But they have raised concerns that some provisions in these bills may be overly broad or difficult to implement.”

Saunders indicated that her organization would consider supporting the bills with further amendments, to “ensure the standards are practical, and avoid unintentionally restricting responsible uses of AI.”

The biggest AI policy discussion at the Capitol is still to come when lawmakers try to make changes to Colorado’s first-in-the-nation AI anti-discrimination law that is scheduled to go into effect in June. It sets rules for businesses and governments that use AI systems in some of their key decision-making to ensure transparency and prevent bias.

The business community has deep concerns about implementation and legal liability, and the Trump administration is also pushing back.

Gov. Jared Polis has created a task force that is negotiating potential changes to the law.