Colorado HB26-1139: No Reimbursement for AI Psychotherapy | TLY

AI Regulation Tracker  /  Regulation signed

Colorado stops carriers paying for AI-delivered psychotherapy from January 1

HB26-1139 does two things at once. It requires a licensed clinician, physician, or other competent regulated professional to review any denial of coverage for a service based in whole or in part on medical necessity when an AI system contributed to the determination, and it forces disclosure of where AI sits in utilization review. It also bars carriers and state health programs from reimbursing psychotherapy conducted by an AI system. Signed June 2, 2026, effective January 1, 2027.

Colorado HB26-1139 use of artificial intelligence in health care regulation briefing
The Leveraged Years AI Regulation Tracker

Colorado now regulates artificial intelligence in health care from both ends. In June 2026 the governor signed two separate acts a day apart. HB26-1195, signed June 3 and effective August 12, 2026, governs what a licensed mental health professional may do with AI inside their own practice. HB26-1139, signed June 2 and effective January 1, 2027, governs what payers may do with AI when deciding whether to cover care, and what they may not pay for at all.

The two were not merged. They address different actors and impose different duties, and a practice or plan operating in Colorado has to satisfy both.

How the bill narrowed

HB26-1139 was introduced on February 4, 2026 and assigned to House Health and Human Services. It was amended in committee on March 4, passed second reading with amendments on March 13, and cleared third reading 47 to 15 on March 16. It moved to the Senate on May 4, was referred unamended by the Business, Labor and Technology committee on May 5, and passed the Senate 32 to 0 on May 11. The Speaker and Senate President signed on May 28, and the governor signed on June 2, 2026. Prime sponsors were Representatives Junie Joseph and Sheila Lieder and Senators Lisa Cutter and Lindsey Daugherty.

The 32 to 0 Senate vote is notable, as is what the bill lost along the way. Earlier versions included provisions addressing mental health chatbots and the unauthorized practice question. Those provisions were removed during the legislative process. The signed act is confined to utilization review requirements and the psychotherapy reimbursement prohibition, which is why it lands as a payer statute rather than a provider one.

The utilization review conditions

From January 1, 2027, an entity using an AI system in utilization review must ensure that determinations account for the patient's medical or clinical history, the patient's individual clinical circumstances, and other relevant clinical factors, as specified in HB26-1139. A carrier's denial of coverage based in whole or in part on medical necessity may not be issued solely on the output of an AI system without human review by a licensed clinician or physician or other competent regulated professional.

The act also specifies that AI systems may not rely solely on group data and must comply with applicable nondiscrimination law. It requires periodic audits of AI systems for accuracy.

The phrase "in whole or in part" is the operative one for compliance teams. A denial does not escape the human review requirement because a clinician contributed some judgment alongside the model. If medical necessity is part of the basis and an AI system contributed to it, the licensed reviewer requirement attaches.

The disclosure obligation is the new administrative work

The provision that will consume the most implementation time is the reporting duty. Entities using AI systems must disclose to the Division of Insurance, the Department of Human Services, or the Department of Health Care Policy and Financing, as applicable, three things: the utilization review functions for which the AI system will be used and the points in the utilization review process when it will be used, the process for human oversight of adverse coverage determinations, and the process for maintaining audit information.

This is a structural map of the review pipeline handed to a regulator, not a general assurance. It requires a payer to know, and to write down, exactly where in the sequence a model contributes. Organisations that have layered vendor tools into utilization review incrementally may find that assembling this document is the hardest part of compliance.

The reimbursement ban

The act prohibits carriers and payers under the Colorado Medical Assistance Act and the Children's Basic Health Plan from reimbursing psychotherapy services that are provided directly to a client and that are conducted by an AI system. Carriers, and payers under the Colorado Medical Assistance Act and the Children's Basic Health Plan, are prohibited from reimbursing psychotherapy services that are provided directly to a client and that are conducted by an AI system.

The prohibition targets the payment channel rather than the product. The act prohibits carriers and state health programs from reimbursing psychotherapy services that are provided directly to a client and conducted by an AI system. The prohibition targets the payment channel, not the product's existence. For a digital mental health company relying on Colorado reimbursement, the prohibition removes the payment channel for psychotherapy conducted by an AI system directly with a client.

HB26-1195, signed June 3 and effective August 12, 2026, governs licensed mental health professionals' use of AI in their own practice. HB26-1139, signed June 2 and effective January 1, 2027, governs payers' use of AI in utilization review and reimbursement. The two acts are separate and complementary, targeting different actors.

Where this fits in the 2026 payer wave

Colorado is one of several states that legislated on AI in health coverage during 2026, and the effective dates are staggered across two years. Washington SB 5395 applied from June 11, 2026 and Iowa HF 2635 from July 1, 2026. Alabama SB63 applies from October 1, 2026 and requires a licensed professional on every denial, delay, or modification. Colorado HB26-1139, Utah SB 319, and Georgia SB 444 all apply from January 1, 2027. Illinois SB3114, the Transparency in Downcoding Act, follows on January 1, 2028.

The common core across these statutes is that AI may assist but a licensed human must decide an adverse determination. The variation is in the disclosure architecture. Alabama runs its transparency through an annual certification to the Department of Insurance. Colorado runs it through a process map filed with whichever division regulates the line of business, plus audit retention. A multi-state payer needs both artifacts, and they are not the same document.

Questions this raises

Does Colorado HB26-1139 ban AI in utilization review?

No. It permits AI but conditions it. Determinations must account for the patient's own clinical history and circumstances, systems may not rely solely on group data, and a medical necessity denial based in whole or in part on AI output requires human review by a licensed clinician or other competent regulated professional.

What exactly is prohibited from reimbursement?

Psychotherapy services that are provided directly to a client and that are conducted by an AI system. The prohibition applies to carriers and to payers under the Colorado Medical Assistance Act and the Children's Basic Health Plan.

What has to be disclosed, and to whom?

Entities must disclose to the Division of Insurance, the Department of Human Services, or the Department of Health Care Policy and Financing, as applicable, the utilization review functions and process points where AI is used, the human oversight process for adverse coverage determinations, and the process for maintaining audit information.

How does this relate to Colorado HB26-1195?

They are separate and complementary. HB26-1195, effective August 12, 2026, governs licensed mental health professionals using AI in their own practice. HB26-1139, effective January 1, 2027, governs payers. They were not merged during the legislative process.

Sponsored Training

Browse the full AI Regulation Tracker →

Drafted with AI assistance and verified against the primary source by a human editor. Informational only, not legal advice.