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

Medicare’s new WISeR model, an AI‑driven prior‑authorization system, has just been exposed through roughly 1,000 pages of CMS records released after a Freedom of Information Act suit by the EFF. The documents reveal a cascade of delays, denials, and even patient harm, raising urgent questions about government oversight, vendor incentives, and the rights of seniors relying on federal health coverage.

Transparency Gap in Medicare’s AI Prior Authorization

The EFF’s March lawsuit forced CMS to disclose contracts, internal reports, and provider complaints that had been hidden from public view. Approximately 1,000 pages of records now show how the WISeR model was rolled out in six states without clear safeguards or public explanation. This opacity prevents beneficiaries, clinicians, and watchdogs from assessing whether the AI tools meet legal standards for fairness and accountability.

Without transparent algorithmic documentation, the agency cannot demonstrate compliance with existing health‑information statutes such as HIPAA, nor can it prove that the AI does not embed bias. The newly released internal status reports describe “operational chaos” and “widespread delays,” yet no corrective measures are outlined, suggesting a systemic failure to meet basic procedural due process.

Financial Incentives and Vendor Conflict of Interest

CMS contracts with private tech firms pay vendors based on “averted expenditures,” effectively rewarding them for denying care. Payment tied to cost avoidance creates a direct financial motive to flag more requests as unnecessary, regardless of clinical nuance. This structure mirrors a perverse incentive model that can erode trust in public health programs.

The records reveal that vendors receive “performance bonuses” linked to the volume of denials, a detail absent from public procurement summaries. Such arrangements clash with the principle that government contracts should prioritize patient outcomes over cost savings, and they may run afoul of anti‑kickback regulations designed to prevent profit‑driven interference in medical decision‑making.

Clinical Impact and Potential Discrimination

Providers report that AI recommendations, though supposedly advisory, heavily influence the final human clinician review. Research cited in the documents confirms that “AI‑generated recommendations often influence human decisions,” meaning the system can propagate hidden biases into real‑world care. This raises civil‑rights concerns, especially for marginalized groups who already face disparities in access to treatment.

Patients in the six pilot states have experienced “improper denials, administrative friction, and lengthy delays,” with some left “waiting in pain.” The lack of an independent audit trail makes it impossible to determine whether these outcomes disproportionately affect certain demographics, a gap that could trigger future litigation under the Americans with Disabilities Act or the Equal Protection Clause.

What This Actually Means For You

1. Expect longer wait times for Medicare‑covered procedures in states where WISeR is active, as AI‑driven checks add an extra layer of approval.
2. Monitor your provider’s communications for mentions of “prior‑authorization denials” that may be AI‑generated, and be prepared to request a manual review.
3. Advocate for transparency by contacting your congressional representative and demanding that CMS publish the WISeR algorithmic criteria and vendor contracts.
4. Stay informed about potential discrimination claims; if you belong to a protected class, consider consulting a health‑rights attorney if you suspect bias.
5. Support organizations like the EFF that use FOIA litigation to expose hidden government programs, as their work can lead to policy reforms.

Immediate Action Steps

First, request a copy of your own prior‑authorization decision from your Medicare provider, specifying whether an AI system was involved. Second, if you encounter an unexpected denial, invoke the right to a “qualified human clinician” review as mandated by CMS, and document the interaction for possible escalation.

Frequently Asked Questions

What is the WISeR model and why does it matter?

WISeR is a Medicare pilot launched in January 2026 that uses AI to evaluate prior‑authorization requests in six states. The model matters because the released records show it has caused delays, denials, and possible patient harm.

How does the vendor payment structure affect care decisions?

Vendors are compensated for “averted expenditures,” meaning they earn more when they deny services. This financial incentive can bias the AI toward rejecting claims, undermining patient access.

Can patients challenge AI‑driven denials?

Yes; CMS requires that a qualified human clinician review all denials, and patients can request this review. Documenting the denial and asking for a manual reassessment are essential first steps.

What Do You Think?

Given the clear conflict between cost‑saving incentives and patient rights, should Medicare abandon AI‑driven prior authorization until robust oversight mechanisms are in place?