Six states now impose disclosure requirements for AI use in clinical or coverage decisions, and no two define the trigger the same way. Some require disclosure when a model influences a decision, some when it makes one, and at least one when it is merely available to the decision-maker.

For a health system operating in one state, this is a compliance question. For a system operating in nine, it is an architecture question, and the answer nearly all of them have reached is to implement the strictest standard everywhere rather than maintain state-specific logic in clinical workflows.

The practical effect is that the strictest state legislature is setting national policy for large operators, which is a familiar dynamic in other regulated industries and is not obviously the outcome the other five legislatures intended.

Federal preemption has been floated and has few enthusiastic supporters. Health systems mostly want uniformity rather than any particular standard; patient advocacy groups are wary of a federal floor that would displace stronger state rules. That disagreement is likely to keep the current patchwork in place for some time.