On January 1, 2026, the Centers for Medicare & Medicaid Services switched on a program with an acronym that sounds almost cheerful: WISeR, for Wasteful and Inappropriate Service Reduction. The pilot lets AI systems screen Medicare prior-authorization requests in six states — Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington — before a human ever looks at the file. CMS frames it as a way to catch unnecessary procedures faster. Physician groups frame it as the latest version of a fight that has been building for years: whether software should get a vote on whether a patient’s treatment gets paid for.
A problem years in the making
Prior authorization was already a sore subject before AI entered the picture. According to KFF, Medicare Advantage insurers processed nearly 53 million prior-authorization requests in 2024 and denied 4.1 million of them, a 7.7 percent denial rate. Only 11.5 percent of those denials were ever appealed — but when patients or providers did appeal, 80.7 percent of the denials were fully or partially overturned, a gap that critics say shows the initial denials were often wrong in the first place. A Commonwealth Fund survey found that 41 percent of people who experienced a denial said it delayed their care, and more than a quarter said their condition got worse while they waited.
What WISeR actually does
The WISeR model doesn’t replace human reviewers entirely; CMS and a coalition of insurers have pledged that licensed clinicians will review all clinical denials rather than letting an algorithm issue a final no. But the AI layer decides which claims get flagged for scrutiny and how quickly, which is precisely what worries critics. A widely cited lawsuit over naviHealth’s “nH Predict” tool alleged that the software evaluated post-acute care cases using population-level statistical patterns rather than an individualized review of the patient in front of it, pushing toward denials that were later overturned on appeal far more often than human-only decisions.
Doctors are already uneasy
A survey from the American Medical Association found that 61 percent of physicians worry AI will worsen denials of treatments they consider medically necessary. That skepticism is notable because the AMA has otherwise been receptive to AI tools that cut down documentation time. The difference, physician groups argue, is between AI that helps a doctor work faster and AI that stands between a doctor’s order and a patient’s care.
States are moving faster than Washington
Frustration with algorithmic denials has already produced legislation. California’s SB 1120, which took effect this year, requires that a licensed physician — not software alone — review any AI-assisted denial of care, and it requires that physician to have expertise relevant to the patient’s condition. At the federal level, new CMS rules (CMS-0057-F) require payers to issue prior-authorization decisions within 72 hours for expedited requests and seven calendar days for standard ones starting this year, a response to years of complaints about requests sitting in limbo for weeks.
What supporters say is different this time
CMS and insurer groups argue WISeR is meant to speed up approvals, not just flag denials — catching clearly unnecessary services so reviewers can focus their attention on borderline cases, and theoretically getting legitimate claims through faster. Supporters note that the program is a pilot limited to six states and a defined set of services, with built-in human review requirements that earlier controversial tools reportedly lacked.
What’s next
The WISeR pilot runs through 2026 with an eye toward a possible nationwide expansion, which is exactly what has patient advocates nervous — a small, closely watched trial can look very different once it scales. Watch for three things in the coming months: whether CMS publishes denial-rate data comparing AI-flagged and human-only reviews, whether more states follow California’s lead on mandatory physician review, and whether the appeals-overturn-rate gap narrows or widens. If WISeR’s denial rates end up tracking the pre-AI baseline, insurers will call it proof the system works as designed. If they spike, the pressure for federal legislation mirroring SB 1120 will be difficult for CMS to ignore.
Photo: https://kaboompics.com/ / PEXELS via Pexels
