HB1139 regulates the use of artificial intelligence in Colorado health care, with a focus on utilization review, coverage decisions, and psychotherapy services. The bill declares that health-care decisions should remain human-centered and that AI may be used as an assistive tool, but not as a substitute for clinical judgment in decisions affecting access to care. It applies to carriers, pharmacy benefit managers, private utilization review organizations, behavioral health administrative services organizations, and managed care entities that use AI in utilization review.
Under the bill, if AI is used in utilization review, the entity must ensure the system relies on an individual patient’s clinical history and circumstances, not just group data; is not used in a discriminatory manner; is fairly and equitably applied; and maintains documentation, audit logs, and governance records. The bill also requires written disclosures to state regulators describing how AI is used, where in the review process it is used, and what human oversight exists. Most importantly, a denial of coverage based in whole or part on medical necessity cannot be issued solely on AI output; it must be reviewed and approved by a qualified human clinician, physician, or other regulated professional.
The bill also prohibits health plans and public payers from covering psychotherapy services that are delivered directly to a patient by an artificial intelligence system. It clarifies that ordinary software tools such as billing systems, electronic health records, video platforms, and messaging tools are not themselves considered AI-delivered psychotherapy when used incidentally to human-provided care. The prohibition applies to commercial health benefit plans, Medicaid-related coverage under the Colorado Medical Assistance Act, and the Children’s Basic Health Plan.
The bill’s impact is to add new statutory requirements in Title 10 for private health insurance utilization review and in Title 25.5 for public health-care payment rules, while also tying the definition of artificial intelligence to Colorado’s broader AI statute. It creates compliance obligations for insurers and managed care organizations, limits automated denial practices, and bars payment for AI-only psychotherapy. The effective date is January 1, 2027.
Overall sentiment appears supportive but not unanimous. The bill passed the House Health & Human Services Committee with amendments on strong votes, passed House third reading 44-14, and later passed the Senate Business, Labor, & Technology Committee unanimously before Senate third reading 31-3. The main point of contention is the degree to which AI should be allowed to influence coverage and utilization review: supporters emphasize patient safety, transparency, and human accountability, while opponents likely object to the added regulatory burden or to restricting AI-assisted efficiency in health-care administration.
HB1139 amends Colorado insurance and public benefits law by adding new restrictions on AI use in utilization review and by prohibiting payment for psychotherapy services delivered directly by AI. It affects carriers, pharmacy benefit managers, private utilization review organizations, behavioral health administrative services organizations, managed care entities, Medicaid payers, and Child Health Plan Plus-related payers. The bill also creates disclosure, documentation, and human-review requirements that will shape insurer and managed-care compliance practices beginning January 1, 2027.
The bill’s legislative history suggests generally favorable sentiment, with broad bipartisan support at later stages and unanimous committee approval in the Senate business and technology committee. The House floor vote was more divided but still comfortably in favor. The discussion reflected a policy preference for keeping health-care decisions human-centered and limiting automated denials, while still allowing AI as a support tool for administrative efficiency.
The central contention is whether AI should be permitted to materially influence adverse coverage decisions without human intervention. Supporters of the bill argue that medical necessity denials and psychotherapy should remain under the control of licensed clinicians because AI cannot account for individual circumstances, bias risks, or the therapeutic relationship. Potential critics are likely to focus on operational costs, reduced automation, and the possibility that the bill constrains innovation or efficiency in utilization review and behavioral health delivery. Another point of tension is the bill’s broad reach across private insurance, Medicaid, and behavioral health systems.