States Tighten Rules on AI in Health Insurance Decisions
Six state legislatures have enacted or are considering bills that restrict how insurers use artificial intelligence in prior authorization and utilization review. The measures require human oversight of AI-generated determinations and mandate disclosure of AI use to patients and providers.
Pennsylvania's House Bill 1925 and Senate Bill 1113 prohibit AI from overriding clinical judgment. A healthcare provider must review medical records, document the assessment, and exercise independent judgment before any adverse determination takes effect. Insurers must file annual compliance statements with the state insurance department and disclose AI use to providers and members.
Varying Standards Across States
The proposed legislation differs in how strictly it constrains AI. Oklahoma's House Bill 3675 allows AI to make coverage determinations if a qualified human professional reviews the decision first - but does not require that person to exercise independent judgment or reject the AI's recommendation. The bill imposes no disclosure requirement to patients or providers.
Indiana's approach, enacted March 4, targets a specific problem: downcoding claims based on medical necessity. The law prohibits insurers from using AI as the sole basis for downcoding unless a human first reviews the medical record. It also restricts providers from submitting claims via AI without human review.
Louisiana's Senate Bill 246 takes a stricter position. AI may only assist in the process. A licensed physician must review all medical records and sign any adverse determination. If a patient appeals based on AI use, insurers cannot use AI in the subsequent review.
New Hampshire's House Bill 1406 requires insurers to maintain written records of AI use and document whether AI assisted in claim reviews. An adverse determination must be made by a qualified healthcare provider, with written notice to members and providers explaining the decision.
Compliance and Oversight Requirements
Alabama's Senate Bill 63, enacted April 17, requires insurers to certify that AI tools are reviewed regularly for accuracy and applied fairly across subscriber groups. Insurers must disclose AI use prominently to policyholders.
Four states - Pennsylvania, Louisiana, Alabama, and New Hampshire - require insurers to disclose AI use to patients or providers. Louisiana and Alabama give state insurance commissioners authority to audit and inspect AI systems.
The variation across states reflects ongoing disagreement about how much human judgment AI decisions require. Some bills demand independent review; others require only that a human sign off on what the AI recommends.
For insurance professionals, these requirements mean building new compliance infrastructure: tracking AI use in determinations, training staff on documentation standards, and preparing disclosure materials for members and providers. States that haven't yet acted may follow similar patterns.
Learn more about AI for Insurance and AI for Healthcare.
Your membership also unlocks:
