States are circling common regulatory approaches for AI in prior authorization

What do deep red Texas and bright blue Maryland lawmakers have in common? Not much, generally. But when it comes to regulating the use of artificial intelligence in prior authorization and claims reviews, there are more similarities than differences in their approaches.  

Both states, along with a handful of others, are implementing laws and regulations with shared themes regarding the utilization of AI tools in decisions about treatment approvals and coverage determinations.  

An interesting pattern is emerging as they do so: no matter their partisan affiliation, lawmakers are agreeing that AI tools cannot be left on their own to make important choices about who gets what care.   

Instead, state governments are codifying the place of humans in the loop and establishing similar guardrails to ensure that algorithms work as intended within a safe and transparent process. 

The growing role of AI in prior authorizations and claim determinations

 Prior authorizations (PAs) are widely considered one of the most burdensome administrative tasks in healthcare, siphoning an average of 13 hours a week from practices. They have been associated with delays in care, adverse events and outcomes, and increased levels of burnout for physicians, the American Medical Association says. 

Yet payers rely heavily on PAs as a method of controlling costs and ensuring adherence to standards of care. That being said, many of the nation’s largest payers do recognize the untenable burden of current PA processes on their provider partners, and are working on ways to simplify the submission and adjudication of requests and claims. 

Increasingly, this means employing artificial intelligence tools to make their internal processing faster and more efficient so that providers get answers quickly with fewer steps. 

However, the adoption of AI for these purposes has already been fraught with problems.  Major lawsuits and journalistic investigations have alleged that some payers are using AI to improperly filter, adjudicate, and deny requests for services,  which raises a host of new questions about the role of algorithms in vital care decisions. 

While CMS has taken some initial steps to clarify how AI can and should be used for decision-making in the Medicare Advantage space, the Trump Administration has generally preferred a hands-off approach to regulating AI in healthcare and elsewhere.  

As a result, state legislatures are shouldering the work of defining AI’s place in the PA environment. 

How state policies on prior authorization are shaping up

State-level insurance regulators have recognized the potential issues around AI use in utilization review for some time, and at least half of states have issued some type of guidance around the topic.  

However, as of April 2026, only nine states have laws specifically pertaining to the use of AI in coverage determinations on their books, says a new report from KFF.   

Washington, California, Utah, Nebraska, Texas, Illinois, Indiana, Maryland, and Alabama are leading the way in establishing the parameters for payers operating in the regions, and are therefore helping to define what may become a broader national consensus on this type of AI adoption.  

While each state’s laws include requirements that are specific to their communities, KFF notes that there are more shared ideas than otherwise in the way lawmakers are establishing their guardrails. 

The top priority for state governments is keeping humans deeply in the loop of decision-making. Not only do some of these states require that AI tools cannot be the sole arbiter of a care decision, but others say that only licensed care providers can issue denials.  

In Illinois, for example, the law says that even when an algorithm is used in the course of a utilization review for medical necessity, the health plan “shall ensure that only a clinical peer makes any adverse determination based on medical necessity.”  The law also includes the restriction that only a clinical peer may review an appeal.  

Human supervision of algorithms for accuracy, reliability, and potential bias is also a shared value across states. California requires periodic review of algorithms to ensure they are functioning appropriately, while Texas says a commissioner is allowed to “audit and inspect a utilization review agent’s use of an automated decision system for utilization review at any time.”  

In several states, transparency requirements extend to disclosing the use of AI in utilization review to enrollees, healthcare providers, and state entities. Other themes include mandating that AI tools be applied “fairly and equitably” to prevent bias, and the implementation of limits on how patient data is used for decision-making to protect patient privacy. 

Will the federal approach throttle state regulations before they mature?

The rise in legislative action indicates that concerns about AI in prior authorizations are shared across borders, and likely affect a large number of the states that haven’t yet taken a vote on official measures, as well. 

But actions by the federal government to stifle state-level lawmaking around AI may result in lost momentum just as interest in the topic starts to surge. 

The Trump Administration has repeatedly tried to prevent states from making any of their own decisions about AI legislation, first with a failed attempt at a provision tucked within the national budget and later via an executive order and a national framework that calls for Congress to override state laws with federal standards for AI use.  

These activities could have a chilling effect on state governments that aren’t interested in picking a fight with the White House on AI regulation, and leave gaps in governance that could have a negative impact on providers and patients. 

However, since many states seem to agree on the basic principles of how AI should and shouldn’t be applied to utilization decisions, there is an opportunity for Congress to come together and develop a national framework that may be even more effective and easy to implement than a patchwork of individual state laws. 

The challenge will be finding the balance between guidelines that work on a federal level while still allowing states to develop regulations that are tailored to their specific needs to protect patients and improve the efficiency and accuracy of the PA process. 

 Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].