Medicare's AI-driven prior authorization program is experiencing widespread operational failures just months after launch, according to newly obtained records from a lawsuit filed by the Electronic Frontier Foundation. The documents reveal that the Centers for Medicare & Medicaid Services (CMS) issued requests for corrective action to at least four of six vendors participating in the WISeR (Wasteful and Inappropriate Service Reduction) program within three months of its 2026 launch.

Rapid Deployment, Immediate Failures

The WISeR program launched just six months after its announcement, a timeline CMS itself acknowledged as "extremely condensed." According to over 800 pages of released records, the accelerated deployment appears to have compromised quality and oversight, with CMS sending Requests for Information (RFIs) or Corrective Action Plans (CAPs) to the majority of participating vendors shortly after operations began.

Three vendors faced particularly serious issues. Virtix was cited for determination turnaround times exceeding guidelines, potentially inaccurate denials, and inadequate customer service. Zyter, which operates using Palantir infrastructure, failed to issue determination letters in a timely manner and lacked adequate processes for addressing provider concerns. Innovaccer received two CAP requests and a "Performance Notice of Non-Compliance" for portal functionality failures that directly impacted care coordination.

Troubling Incentive Structure

The documents show that CMS acknowledged WISeR's payment structure gives vendors "an incentive to deny as many claims as possible." This admission aligns with information from vendor applications and CMS inquiries indicating that at least one vendor denied more prior authorization requests than it approved during the program's initial months, with CMS inquiring about "denials that appear inaccurate or inconsistent with applicable criteria."

The records also expose broader concerns about the quality of human oversight in AI-driven healthcare decisions. While CMS requires "human clinician" review for non-affirmations, the documents cite at least one health insurance company's AI system as an example of human-in-the-loop limitations: that system allowed reviewers to spend an average of just 1.2 seconds per case, raising questions about whether human oversight serves as a meaningful safeguard or merely rubber-stamps algorithmic decisions.

Impact on Providers and Patients

Healthcare providers documented in the records face unclear escalation processes and inadequate vendor communication during time-sensitive care decisions, directly impacting their ability to deliver timely treatment. For Medicare beneficiaries, the failures translate into potential care delays and inappropriate denials driven by AI systems deployed without adequate testing periods or legal safeguards.

The document release included applications from 66 companies that sought to participate in the program, with only six ultimately selected. Vendor contracts and CMS correspondence provide insight into how Medicare is outsourcing clinical decision-making to private AI companies, though critical details about training data, decision algorithms, and the specific implementation of human oversight remain unclear.

The EFF has urged transparency, oversight, and legal limits to protect people from the harms of automated decision-making in high-risk systems, arguing that Congress should place greater legal limits on the use of automated decision-making in Medicare.

Sources

  1. New Records Obtained by EFF Expose More Problems with Medicare's AI Experiment — Electronic Frontier Foundation