Documents released in mid-September 2026 show that the Centers for Medicare & Medicaid Services’ first large-scale experiment in AI-driven prior authorization has produced a wave of denials, backlogs and missed deadlines just months after launch. The records, obtained through public disclosure requests and reported by STAT News, cover the first three months of the Wasteful and Inappropriate Service Reduction model, known as WISeR, which CMS rolled out in January 2026 across six states.
What WISeR Was Supposed to Do
CMS designed WISeR to target specific procedures it considers vulnerable to fraud, waste and abuse in traditional fee-for-service Medicare, including nerve stimulator implants and steroid injections for chronic pain. Rather than requiring prior authorization for all Medicare enrollees, the agency contracted with private technology vendors to build AI systems that would flag and review a narrower set of services in Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington. The pitch was efficiency: faster reviews, fewer unnecessary procedures, and lower costs to the program, all without adding new bureaucracy for beneficiaries in Original Medicare.
What the Numbers Actually Show
The newly disclosed records tell a different story. Two vendors together denied more than 20,000 prior-authorization requests during the pilot’s first three months. One contractor, Virtix, made more than 6,000 decisions in that window and denied 53% of them — meaning it rejected more requests than it approved. That denial rate was steep enough that CMS required Virtix to submit a corrective action plan. A separate tally cited in the documents shows the two vendors combined for 5,944 denials in a subset of the reviewed period, underscoring how concentrated the rejections were among a small number of service categories.
Deadlines Came and Went
Beyond the denial rates, the records expose operational breakdowns. CMS publicly states that vendors should respond to prior-authorization requests within 72 hours. The documents show that standard was routinely missed, with one request going unanswered for 83 days — more than a month beyond even a generous reading of “expedited” review. Patient advocates and provider groups say delays of that length can mean seniors go without needed nerve stimulators or pain treatment for months while paperwork sits with an AI vendor’s review queue.
Two Very Different Readings of the Same Data
CMS has defended WISeR as a necessary check on wasteful spending, arguing that some of the flagged procedures have historically been over-utilized in ways that cost the program money without improving outcomes. Vendors participating in the pilot have said early-stage friction is expected as clinical rule sets and provider documentation practices are tuned to the AI review process. But physician groups, including voices cited in the American Medical Association’s ongoing surveys on payer AI, argue the pattern looks familiar: AI systems calibrated more for claims reduction than clinical nuance, producing denials that get overturned on appeal at high rates once a human reviewer looks at the case. The National Health Law Program and the Electronic Frontier Foundation have both flagged the WISeR records as evidence that automated review systems are being deployed in Medicare before adequate testing and oversight structures are in place.
A Pilot Under a Bigger Spotlight
The disclosures land amid a broader national reckoning over AI’s role in coverage decisions. UnitedHealth Group, Humana and Cigna are all facing active lawsuits alleging their AI tools improperly denied Medicare Advantage claims, and several states have passed laws in 2026 requiring insurers to disclose when AI is used in medical-necessity reviews. WISeR is notable because it is a federal government program, not a private insurer’s internal tool — meaning CMS itself is now the subject of the same criticism it has leveled at commercial payers.
What Happens Next
CMS has not announced plans to pause or expand WISeR beyond the six pilot states, but the agency faces pressure from both directions: fiscal hawks who want the program to prove it can cut wasteful spending, and lawmakers and advocacy groups pushing for stricter turnaround-time enforcement and denial-rate transparency. A Senate vote earlier this year left the pilot intact, but the newly public numbers give critics fresh ammunition ahead of the program’s next review milestone. For now, providers in the six affected states say they are advising patients to expect delays and to be ready to appeal denials that come back with little clinical explanation — a workaround that undercuts the very efficiency gains WISeR was built to deliver.
Photo: com329329 / PIXABAY via Pixabay
