As artificial intelligence (“AI”) continues to be deployed in new use cases, state legislatures are considering measures to oversee its use in health insurance, particularly in prior authorization and utilization review. Last year, legislatures in Arizona, Maryland, Nebraska, and Texas enacted bills to address AI use in health insurance determinations.[1] A prior analysis of California legislation that aimed to regulate AI in utilization review in the managed care space is available here. Building on that work, this post summarizes legislation under consideration in the first quarter of 2026 that governs health insurers’ use of AI in determinations such as prior authorization and utilization review.
Pennsylvania
In Pennsylvania, the legislature is considering nearly identical proposals in both chambers – House Bill 1925[2] and Senate Bill 1113.[3] Under both bills, insurers may use AI in utilization review, but AI-based algorithms may not override the clinical judgment of healthcare providers, and determinations must be grounded in (1) clinical history, (2) the individual clinical or nonclinical circumstances identified by the requesting provider, and (3) other relevant information in the enrollee’s medical record. Before issuing any determination, a healthcare provider must review the enrollee’s clinical records and other pertinent information, document the assessment, and exercise judgment independent of AI recommendations when making any adverse benefit determination.
Insurers that use AI in utilization review would be required to file an annual AI compliance statement with the Pennsylvania Insurance Department, and to disclose their use of AI in utilization review to providers and members. While the bills are nearly identical, Senate Bill 1113 additionally authorizes the temporary suspension of new enrollments by insurers that fail to comply with the bill’s requirements.
New Hampshire
New Hampshire House Bill 1406[4] would permit insurers to use AI as an assistive tool in utilization review, and would require insurers to maintain written records related to AI use, including “determinations affecting provider coding decisions.” Similar to other proposals discussed here, the bill provides that adverse determinations must be made by a “qualified healthcare provider.” If enacted, insurers would need to provide written notice of an adverse determination to members and providers detailing the basis for the decision, and to document in their records whether AI was used to assist in the review of claims.
Oklahoma
If signed into law, Oklahoma House Bill 3675[5] would allow insurers to use AI in making coverage determinations, but would prohibit AI from serving as a final decision-maker without prior review by a “qualified human professional.” Although the bill requires qualified human professionals to review each determination, it does not require them to exercise independent judgment or prohibit their reliance on the AI’s output.
House Bill 3675 does not impose explicit documentation requirements or mandate submission of an annual compliance statement. It does, however, authorize Oklahoma’s insurance commissioner to audit and inspect use of AI in utilization review at any time, effectively necessitating that insurers maintain records necessary to comply with this obligation. Unlike other legislation evaluated here, House Bill 3675 would impose no obligation on insurers to disclose AI use to members and providers.
Indiana
Enacted on March 4, 2026, Indiana House Bill 1271[6] establishes a new chapter governing the downcoding of health benefits claims (IC 27-1-52), with provisions directly addressing AI use in health insurance determinations. The bill prohibits insurers from using any automated process, system, or tool — including artificial intelligence — as the sole basis to downcode a claim based on medical necessity unless a human employee or contractor has first reviewed the covered individual’s medical record. Notably, the bill imposes a parallel obligation on providers, prohibiting the use of AI to submit a health benefits claim without review by a provider or other person involved in developing the claim. Insurers are also required to disclose, in an easily accessible and readable manner, whenever AI is used to make an adverse determination on a prior authorization request or to downcode a claim.
Louisiana
In Senate Bill 246,[7] the Louisiana legislature proposes strict limitations on the use of AI by defining the role AI should play in the adverse determination process, and by clarifying the responsibilities of healthcare providers during this process. Although the bill allows use of AI, Senate Bill 246 bars insurers from replacing healthcare providers with AI or an automated decision system for adverse determinations. Under the proposal, AI may serve only as an assistive tool, and a licensed physician must review all medical records and sign any adverse determination. If a decision is appealed based on the use of AI or an automated decision system, insurers would be prohibited from using AI or an automated system in any subsequent review of a claim.
The bill would also require insurers to document their use of AI in determinations and to disclose that use to enrollees and to the Louisiana Department of Insurance in connection with determinations. If enacted, AI and automated systems would be subject to inspection and audit by the Louisiana insurance commissioner.
Alabama
The Alabama legislature enacted Senate Bill 63[8] on April 17, 2026. This legislation establishes parameters for the use of AI, including standards for the data sets on which AI tools rely, requirements for fair implementation in prior authorization, and a reaffirmation of existing prohibitions on discrimination.
Alabama Senate Bill 63 does not confine ultimate determinations to licensed physicians alone, and imposes conditions for AI use by requiring that determinations “always be made by a licensed physician or other healthcare professional.”[9] Under the law, AI may provide recommendations or conclusions to inform determinations, but its role shall be expressly supplemental, because a licensed physician or other healthcare professional must evaluate any AI-generated recommendations or conclusions in light of the enrollee’s circumstances and the treating provider’s recommendations.
In addition, Senate Bill 63 requires insurers to certify with the Alabama Department of Insurance that any AI tools used “are reviewed on a periodic basis to maximize accuracy and reliability,” and that such tools are configured and applied fairly for each subscriber group and member. Insurers are also required to prominently disclose to policyholders their use of AI in utilization review.
FOOTNOTES
[1] H.B. 2175, 57th Leg., 1st Reg. Sess. (Ariz. 2025); H.B. 820, 2025 Sess. (Md. 2025); L.B. 77, 109th Leg. (Neb. 2025); S.B. 815, 89th Leg., R.S. (Tex. 2025). [2] H.B. 1925, 2025–2026 Sess. (Pa. 2025). [3] S.B. 1113, 2025–2026 Sess. (Pa. 2025). [4] H.B. 1406, 2026 Gen. Ct. (N.H. 2026). [5] H.B. 3675, 60th Leg., 2d Reg. Sess. (Okla. 2026). [6] H.B. 1271, 124th Gen. Assemb., 2d Sess. (Ind. 2026). [7] S.B. 246, 2026 Reg. Sess. (La. 2026). [8] S.B. 63, 2026 Reg. Sess. (Ala. 2026). [9] We further note similarities between Alabama’s Senate Bill 63 and Florida’s House Bill 527, which recently died in the Rules Committee in March. The Florida bill would have permitted the use of AI and machine learning systems in claims processing and decision-making, but would have required that such systems function only as assistive technologies, and also would have prohibited claim payment or denial decisions based solely on AI or machine learning systems. H.B. 527, 2026 Leg., Reg. Sess. (Fla. 2026).Source link
