States Are Passing Laws to Stop AI From Denying Your Health Claims

Your health insurer may be using AI to decide whether your claim gets paid. A growing number of states have decided that's a problem worth legislating.

Indiana enacted the first US law specifically targeting AI-based claim downcoding on March 4, 2026. It takes effect July 1. Under the new statute, health insurers cannot use AI as the sole basis for reducing a submitted billing code without reviewing the patient's actual medical record. The law targets a specific practice: automated systems that cut the payout on a claim by lowering its billing category, without a human reviewing the underlying records.

A wave of state action

Indiana wasn't first to act on AI in health coverage generally. In 2025, four states passed laws restricting AI's role in prior authorization and claims decisions:

Arizona (HB 2175): Requires that health care providers independently review claims and prior authorization requests before an insurer can deny them.

Maryland (HB 820): Requires that any AI tool used in utilization review incorporate the patient's medical history, individual circumstances, and other clinical information. Those tools must be open to inspection and audit by the state.

Nebraska (LB 77): Prohibits AI output from being the sole basis for evaluating medical necessity in order to deny, delay, or modify health care services.

California (SB 1120): Any denial or delay of care based on medical necessity must be reviewed and decided by a licensed physician or qualified health care provider, not an automated system.

In 2026, seven states introduced bills with similar downcoding provisions: California, Connecticut, Illinois, Indiana, Maryland, Missouri, and Oregon.

The numbers behind the legislation

The policy push is driven by documented patterns in how AI-issued denials play out. Studies of AI-driven Medicare Advantage prior authorization denials show high overturn rates when patients appeal, but the vast majority of patients never file an appeal. The practical result: AI-generated errors in claims processing often go unchallenged, not because they're correct, but because most people don't know they can contest a denial or don't have the time to do it.

The legislative response is not aimed at banning AI from health care administration entirely. The common framework across state bills is narrower: AI can assist in flagging and processing claims, but a human has to review before any denial stands.

The broader picture

The pace of state-level AI legislation in health care has accelerated sharply. As of early 2026, 43 states have introduced more than 240 bills addressing AI concerns across industries. Health care is one of the most active areas. With federal AI legislation stalled, states are filling the gap. The result is a patchwork of laws that varies significantly by state, which creates its own compliance challenges for national insurers.

KFF Health News covered the current legislative landscape in depth on April 15. The Manatt Health AI Policy Tracker is updated regularly as bills advance. For context on the broader federal picture, our earlier coverage of the EU AI Act's August enforcement deadline shows how differently European regulators have approached similar problems.