In 2025 alone, doctors and legislators across the country have already introduced more than 100 pieces of legislation meant to reform prior authorization. The cumbersome approval process is a leading cause of burnout among doctors, and changing it is a top priority of the American Medical Association (AMA). According to a 2024 survey, 93% of physicians say that prior authorization has led to patient harm.
Leaders in the health-tech arena believe prior authorization is ripe for an artificial intelligence (AI)-driven transformation. The problem is that most doctors aren’t on board.
In February, the AMA reported that 61% of physicians believe that using AI in prior authorization will increase denial rates and delay patient care. Moreover, medical associations in at least five states have introduced legislation meant to limit AI’s use in patient decisions.
Though innovators promise speed and better access, doctors say that insurers could easily use the technology to make approvals and appeals even more taxing. Medscape Medical News spoke to prior authorization experts, including doctors on the front lines, and they all agree that AI has the potential to change the game — but only if prior auth gets a makeover. If the process isn’t reformed to be more transparent and patient-centric, AI will simply make a flawed system work faster.
About Medscape Data
Medscape continually surveys physicians and other medical professionals about key practice challenges and current issues, creating high-impact analyses. For example, the Medscape Physician and Prior Authorization Report 2024 found that:
- 66% of physicians said they spend slightly more or much more time on prior authorization than they did previously.
- 89% of doctors say too many drugs and services require prior authorization.
- 86% say it interferes with patient care.
- 19% are in favor of greater automation.
Could AI Fix the System?
“The danger of layering AI on top of a broken system is that, while it offers a lot in terms of efficiency, if you introduce more efficiency and speed into that system, you just speed up the rate at which you do bad things,” said Amy Killelea, JD, assistant research professor at Georgetown University’s Center on Health Insurance Reforms, Washington, DC.
It’s unclear how widely AI is already used in prior authorization. A new survey of 93 health insurers released at the National Association of Insurance Commissioners (NAIC) conference in March 2025 reported that 92% use, plan to use, or are exploring AI and machine learning. The preliminary results were not explicit about how health insurers are employing AI.
If they’re using the algorithm to collect relevant data from the medical record more quickly, that’s a use case no one would have a problem with, Killelea said. But algorithms making decisions about claims and denying care is a problem. Cigna and UnitedHealth already face class action lawsuits after investigations showed they used AI to deny claims without doctor oversight.
According to Leslie Lenert, MD, a primary care physician who researches computing at the Medical University of South Carolina in Charleston, companies used rule-based AI to compare claims to insurer criteria. These kinds of if/then algorithms are too fragile for the complex reasoning it takes to evaluate claims, Lenert said. The result was thousands of claims denied at a time without a meaningful doctor review.
The Myth of Anti-Tech Doctors
Doctors are rallying to slow the use of AI in prior authorization — but contrary to popular belief and some headlines, it’s not because they’re anti-technology.
Most doctors are pro-AI. “AI is probably a great way to save time and control cost. We have no problem with AI being involved in the first look” at a claim, said Sarah Lee-Davisson, MD, a general surgeon and an advocate for an Arizona bill regulating AI in prior authorization. She said that the apprehension has more to do with the nebulous prior authorization process.
Even without AI, prior authorization is already a black box. Physicians often get little or no feedback on why a claim is denied and the approval metrics are constantly changing. A KFF analysis of ACA Marketplace plans in 2023 found that the most common reason listed for denial, at 34%, was “other reason not listed.”
Brian Callaghan, MD, a neurologist at the University of Michigan in Ann Arbor, Michigan, recently spent months working to get one of his patients rituximab. The claim was initially denied without a clear reason or a clear option to appeal. By contrast, he never has problems getting the drug for his patients at Veterans Affairs, where there’s only one payer.
“With private insurance, you feel like they are trying to deny as many things as possible,” said Callaghan, who is also lead author on a 2024 paper on using AI to refine prior authorization. Some form of cost containment is essential, but it needs to be easy, he said. He said that having a complex process means fewer things are approved — not because of science or clinical appropriateness, but because the process is challenging.
While payers are heavily regulated in other areas, there are few to no guardrails in place to make sure that prior authorization is based on clinical data and in the patient’s best interest. And payers have a clear conflict of interest, said Lenert.
It's not technology that makes physicians nervous; it’s an imbalance of power, Lenert added. “Insurance companies have unlimited access to resources to deny claims and plenty of incentive to do so,” he said. “And providers often don’t have access to any additional resources [to push back].”
Callaghan agrees. Doctors already put in long hours and have designated staff members handling prior authorization. While he’s a firm believer that AI could help solve some of these problems if prior authorization continues as it is — complicated, ever-changing, and hard to speak to a peer physician — then AI may very well make the process even harder, he said.
Holding Out Hope
Still, doctors hope that if prior authorization is made more transparent, AI could relieve a huge burden. The technology could minimize time spent on paperwork for both sides. And if the first look at a claim is automated, approvals could happen faster.
Start-ups in the space are already building AI-based tools to consolidate disjointed insurance portals, speed up pre-visit insurance verification, determine where prior authorization is needed, and automatically submit electronic requests to appropriate payers.
In the future, more sophisticated generative AI algorithms probably could tackle denials, Lenert said. “Even if [AI could] turn around claim denials faster and give an explicit reason why, that would be a service.” That would give doctors far more information than they have to go on today.
But as long as payers “can say no without taking responsibility for the need to act, that isn’t fair,” Lenert said. That’s true, with or without AI. The industry needs a new standard of denials. “If it isn’t fully transparent, it’s fully wrong,” he added.
A Breaking Point
At the state level, more legislation on prior authorization continues to be introduced, the NAIC reported. “It’s reached a breaking point,” Killelea said. Prior authorization has transitioned from a healthcare industry debate to a public concern. “We are already seeing states jump into the fray.”
The aim is not to eliminate prior authorization but to set up some guardrails so the system can’t be used to delay or deny clinically appropriate care. There’s plenty of AI-specific legislation too. California and Colorado have already passed legislation limiting automated decision-making systems in prior authorization. Texas, Arizona, Connecticut, and Indiana have similar bills in the works.
“We have no problem with AI being involved in the first look [at a claim],” said Lee-Davisson in Arizona, “but if there’s going to be a denial of a test or medicine or surgery, we certainly think it warrants a doctor looking at it.”
Zeke Silva, MD, a private practice interventional radiologist in Houston, is backing a similar bill in Texas. “We accept that prior authorization has some role. We accept that algorithms have some role in patient care,” he said. “But if you’re denying care, it should be by someone trained to practice medicine.”
Donavyn Coffey is a Kentucky-based journalist reporting on healthcare, the environment, and anything that affects the way we eat. She has a master’s degree from NYU’s Arthur L. Carter Journalism Institute, New York City, and a master’s in molecular nutrition from Aarhus University in Aarhus, Denmark. You can see more of her work in Wired, Teen Vogue, Scientific American, and elsewhere.
