A pilot program using AI to review Medicare medical expenses is sparking a debate in the US over 'denial incentives.' On September 25, 2026, Ars Technica published an analysis of how the Trump administration's WISeR program works: the Centers for Medicare & Medicaid Services (CMS) launched the pilot in six states this January, using AI plus human clinical review to issue 'affirmed / not affirmed' decisions on prior-authorization requests for certain traditional Medicare outpatient services — and, crucially, the AI vendors involved take a cut of the money Medicare 'saves.'
What WISeR is: six states, six years, a dozen-plus services
WISeR stands for Wasteful and Inappropriate Service Reduction. Launched January 1, 2026 and scheduled to run through December 2031, it covers Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington. The pilot targets 14 to 17 outpatient services CMS deems vulnerable to overuse, including electrical nerve stimulator implants, epidural steroid injections, percutaneous vertebral augmentation, and knee arthroscopy. In practice, providers can submit a prior-authorization request, and the AI vendor returns an 'affirmation / non-affirmation' determination in about three days; any non-affirmed request must be reviewed by a qualified clinician before it can be finally denied — CMS stresses that the AI itself cannot issue the final 'no.'
The core controversy: vendor revenue tied to 'how much was blocked'
According to KFF Health News, WISeR vendors keep 10% to 20% of 'averted expenditures' as compensation. Critics argue that under this design, vendors' natural incentive is to raise denial rates rather than let legitimate requests through. Doctors interviewed by KFF called the process 'horrendous': requests that used to be approved in about two weeks now take four to eight; one Oklahoma radiologist documented four cases noting the patient had 'no numbness,' yet the requests were denied on the grounds that numbness ruled out the spinal procedure. On July 16, the US Senate voted 46 to 50 against a motion to terminate the pilot, letting it continue. The Congressional Research Service previously called WISeR 'one of the largest expansions of prior authorization in the history of traditional Medicare.'
Health data in third-party hands: the privacy and compliance side
Prior-authorization review means large volumes of claims documents containing health information flow through third-party AI vendors' systems. CMS rules require that final denials be made by licensed clinicians and set penalties for 'inappropriate denials' — but how the algorithms score requests, and which factors carry what weight, is not publicly disclosed. For the seniors enrolled, the questions are concrete: who sees their medical records, how long they are kept, and whether they can be used to train models — with no clear public answers so far. That is why, beyond governance and compliance, the privacy dimension cannot be sidestepped either.
Saving money or treating patients — which comes first
WISeR has been running for only seven months, with more than five years to go. Supporters see the Medicare funds wasted each year; opponents see slower approvals and seniors who cannot afford to wait. The Ars Technica analysis put the sharpest point bluntly: when a contractor's revenue is 'the money it blocks,' it has no incentive to let one more request through — unless the timeliness and appeal-adjustment terms in its performance metrics genuinely bite into its wallet. Whether this pilot ends up as a model for 'AI saving Medicare money' or a cautionary tale of 'algorithmic denials' will come down to three numbers in the coming years: denial rates, appeal success rates, and wait times.
