AI is cutting your claims. A new Indiana law fights back.

When a health plan's software quietly knocks an office visit down to a lower-paying code, the practice absorbs the loss. Starting July 1, Indiana becomes one of the first states to put legal limits on that practice when artificial intelligence is making the call.

Indiana's House Enrolled Act 1271, signed in March and backed by the Indiana State Medical Association, bars health plans from using an automated tool, including AI, as the sole basis for downcoding a claim without a review of the patient's medical record, according to the American Medical Association. The law also requires plans to notify physicians when a claim is downcoded, prohibits downcoding based on a diagnosis alone and guarantees an appeals process that lets practices batch downcoded claims.

Downcoding happens when a payer reduces a submitted claim to a lower-level, lower-paying code, so the practice gets paid for less care than it documented. Claim-editing algorithms increasingly do this automatically, without ever opening the chart. The AMA's position is blunt: it is never acceptable to downcode a claim without a review of the medical record, the group says in its guidance for practices fighting these programs.

Indiana is the leading edge of a wider push. Lawmakers in California, Connecticut, Illinois, Maryland, Missouri and Oregon introduced downcoding bills this year, part of a broader wave of 2026 state laws aimed at how insurers use AI across prior authorization and claims review, according to an analysis from law firm Holland & Knight. Alabama, Utah, Washington, Maryland and Georgia have each enacted measures that, in different ways, require a qualified human, not an algorithm alone, to stand behind an adverse decision.

Regulators are acting without new legislation too. The Maryland Insurance Administration recently found that Cigna improperly downcoded evaluation and management claims from physicians, imposed a penalty and ordered the plan to stop, the AMA reported.

For practices, the immediate work is documentation and vigilance. Tight, specific notes are the strongest defense against an algorithm that downgrades a visit it never read. Watch remittances for patterns of reduced E/M levels, flag them fast and appeal rather than eat the cut. The AMA offers sample appeal letters and downcoding resources for practices that want to push back.

The Indiana law takes effect July 1. With more states weighing similar bills and Congress still on the sidelines, where a practice operates increasingly determines how much protection it has when a payer's AI decides a visit was worth less than the physician billed.

Is your practice seeing claims downcoded? Take our short survey and tell us what you're up against. We'll use the results in our continued coverage.