H.R. 10210, the Doctors Not AI Act of 2026, was introduced in the House on September 1, 2026 by Representative Greg Landsman (D-OH-1), for himself and Representatives Buddy Carter (R-Ga.), Kim Schrier (D-Wash.) and Tom Barrett (R-Mich.). It was referred to the Committee on Energy and Commerce and additionally to Ways and Means and to Education and Workforce. The text captured here is the introduced version as published by Congress.gov.
The bill regulates the use of AI in health-plan claim review. It does so through parallel amendments to three statutes so that the same duty reaches insured coverage, self-funded employer plans and the tax-code enforcement channel: title XXVII of the Public Health Service Act (new section 2719B), the Employee Retirement Income Security Act of 1974 (a new paragraph (3) in section 503), and chapter 100 of the Internal Revenue Code (new section 9815A).
Scope
The operative duty reaches adverse benefit determinations involving clinical judgment, initially and on internal appeal. Approvals are not regulated.
An "adverse benefit determination involving clinical judgment" is defined as one based, in whole or in part, on medical necessity, appropriateness, experimental and investigational or similar exclusions or limits, level of care, health care setting, effectiveness, clinical guidelines, utilization review criteria, other standards requiring evaluation of the enrollee's medical condition or treatment needs, or generally accepted standards of care for those needs.
Definition of an artificial intelligence system
The same definition is inserted into all three statutes: "an engineered or machine-based system, including any algorithm, predictive model, machine learning system, or automated decision-making software, that processes data to generate predictions, classifications, recommendations, or decisions and that is used to materially influence, automate, or issue determinations regarding coverage of health care items or services." The definition turns on the use to which the system is put rather than on any technical property, and reaches rule-based scoring software as readily as a trained model. It may be compared with the definitions collected at Automated Decision-Making Technology (ADMT).
The duty
A plan or issuer must ensure that a covered determination:
- Is not issued by an AI system and is not dictated or determined by the output of such system.
- Is made only by a licensed health care professional acting within the scope of the licence and having training and experience in the item or service at issue.
- Reflects the professional's independent clinical judgment. The professional "shall conduct an independent evaluation of the enrollee's individual medical circumstances and shall not treat any output of an artificial intelligence system as presumptively valid or defer to such output in lieu of independent clinical judgment applying generally accepted standards of care."
- Discloses the AI use in the denial notice, if any: a statement that a system was used, a description of its role in the review process, and the name, professional licence and credentials of the professional who made the determination.
- Is supported by documentation in the administrative record describing the system used, its role, any outputs, scores, recommendations or determinations it generated, and documentation demonstrating compliance with the independent-judgment requirement. Those materials form part of the administrative record and are made available to the enrollee on request.
The third element is the bill's operative limit on assisted review. It does not bar an AI system from the process, but it forbids treating the system's output as presumptively valid and requires an independent evaluation of the individual's circumstances in every case.
Mental-health parity
Each of the three statutes' parity provisions is amended twice — Public Health Service Act section 2726, ERISA section 712, and Internal Revenue Code section 9812. First, a new clause (vi) is added to the non-quantitative treatment limitation comparative-analysis requirement, obliging a plan to state "whether an artificial intelligence system is used in, or materially influences, the design, development, application, or administration of such limitation" and, if so, to supply enough information about the system's function, operation and effects for the Secretary to evaluate them as between mental health or substance use disorder benefits and medical and surgical benefits, both as written and in operation.
Second, a new subsection provides that "the use of an artificial intelligence system in connection with utilization review shall constitute a treatment limitation" for parity purposes. This places AI-assisted utilization review inside an existing comparative-analysis and enforcement regime rather than creating a new one.
Effective date
The amendments apply to plan years beginning on or after January 1 of the first calendar year beginning not less than 12 months after enactment.
Textual note
Section 2(a) of the bill cites the Public Health Service Act definitional section as "42 U.S.C. 30099–91." The codification is 42 U.S.C. 300gg–91. The error appears in the introduced text as published.
Relationships
- related: Healthcare — AI Deployment; Insurance — AI Deployment — the regulated activity.
- related: Automated Decision-Making Technology (ADMT) — the definitional family the bill's AI-system definition belongs to.
- related: AI Liability — the bill assigns responsibility for a determination to a named licensed professional.
- related: Doctors Not AI Act of 2026 (H.R. 10210) — the tracking page for the bill.