New Records Reveal Problems with Medicare’s AI Prior Authorization Experiment

A Washington, D.C.-based nonprofit, the Electronic Frontier Foundation (EFF), recently obtained approximately 1,000 pages of records related to the WISeR (Wasteful and Inappropriate Service Reduction) program from the Centers for Medicare & Medicaid Services (CMS). The WISeR program leverages artificial intelligence to evaluate prior authorization requests for certain medical services, which has led to widespread delays and denials of care, operational chaos, and reports of patient harm across six states.

The EFF's Freedom of Information Act (FOIA) lawsuit aimed to gain transparency into the experimental AI program, which could jeopardize Medicare beneficiaries' access to care. CMS launched the WISeR model in January 2026, mandating medical providers to seek prior authorization before delivering specific treatments.

Private companies contracted by CMS evaluate the requests using AI, though little is known about the AI systems. While CMS states that a qualified human clinician must review all denials, research shows that AI-generated recommendations often influence human decisions. Furthermore, the WISeR program offers financial incentives for vendors to deny care, as they are paid for averted expenditures.

According to the records, WISeR vendors have faced significant challenges, including delayed responses to prior authorization requests and inappropriate denials. CMS claims to safeguard against inappropriate denials by tying vendors' payment rates to quality scores, yet the released data indicates that low quality scores only reduce payments by 5-10%.

The records also suggest that a few WISeR vendors have denied thousands of prior authorization requests, raising concerns about the program's effectiveness and potential harm to patients.