EFF releases new records exposing flaws in Medicare's AI prior authorization pilot
EFF obtained new government records detailing additional problems with Medicare's AI-driven prior authorization program, renewing calls for legal limits on automated decision-making in government.
Summary
The Electronic Frontier Foundation has published more than 800 pages of new records about WISeR, the Wasteful and Inappropriate Service Reduction model, a Medicare pilot in which AI assesses prior authorization requests for some medical services. EFF obtained the material through a lawsuit it filed in March against the government. The release includes requests for information or corrective action plans that the Centers for Medicare and Medicaid Services (CMS) sent to four of the six vendors and the vendors' replies, applications from 66 companies, and contracts and task orders covering evaluation and monitoring. EFF says earlier records already showed delays, denials of care and reports of patient harm.
According to EFF, the documents show a rushed rollout. CMS launched the program six months after announcing it, despite calls from providers to wait, and earlier acknowledged an extremely condensed timeline. In the first three months, CMS wrote to at least four vendors about delays, technical failures and poor communication with providers. One vendor, Virtix, was questioned over slow turnaround, denials that seemed inaccurate and weak customer service. Zyter, which uses Palantir infrastructure, was asked about late determination letters. Innovaccer received two corrective action requests and a notice of non-compliance over a missing portal. Vendors blamed compressed timelines, and one said it had warned CMS it might not be ready.
EFF also argues that safeguards remain opaque. It notes that the payment structure gives vendors, as the government itself put it, an incentive to reject claims wherever possible, and that some safeguard details appear redacted. Nobody can yet tell, it says, whether clinician review is real oversight. EFF maintains that AI should not drive consequential government decisions and urges Congress to set legal limits on automated decision-making in Medicare.