Medicare’s Wasteful and Inappropriate Service Reduction (WISeR) model put AI-assisted prior authorization inside Original Medicare on January 1, 2026, across six states, with the technology companies performing the reviews compensated based on a share of averted expenditures. This analysis maps the model’s first six months against a confidence-scored governance framework: the incentive architecture, the GAO determination that the model notice is a rule, the first contractor corrective action plan, and what the enforcement record signals for every payer that follows.
Every finding in the full paper carries a confidence rating, and published findings are updated as the evidence changes, using the same three-pass method applied to every 5Q engagement.
Enforcement arrives: the Virtix corrective action
In late June 2026, CMS ordered Virtix Health, the model participant for Washington State, to submit a Corrective Action Plan after finding the company out of compliance with the model’s required 72-hour turnaround for prior-authorization and prepayment determinations. The finding followed an audit across five domains: clinical determinations, communications, portal functionality, customer service, and timeliness. CMS now holds biweekly performance-review meetings with the contractor.
Read against the NIST AI Risk Management Framework, the corrective action documents a deployer-ecosystem failure across three functions: third-party oversight that relied on complaint-driven detection rather than proactive monitoring (GOVERN); a 72-hour standard that existed on paper from launch with no published mechanism tying measurement to consequence until the CAP (MEASURE); and post-deployment monitoring retrofitted under federal supervision that ISO/IEC 42001’s operational controls would have required as a condition of go-live (MANAGE).
Root cause, by the Five Whys
The missed turnaround times are the visible problem. In my assessment, the compensation model is the underlying condition.
Update: September 28, 2026
Nine-month records. On September 8, 2026, the Electronic Frontier Foundation released approximately 1,000 pages of CMS records on WISeR obtained through FOIA litigation. The records document missed response times, a quality-score adjustment of 5 to 10 percent of vendor payment, and a vendor’s notice to CMS before launch that its system was not ready. They also correct a widely reported figure: the two vendors with published data denied 5,944 requests in the first three months, not 20,000. Read the nine-month update →
Status note — September 26, 2026
Congressional Review Act vote. On July 16, 2026, the Senate rejected, 46–50, a motion to proceed to S.J.Res.198, the resolution of disapproval of the WISeR prior-authorization rule (Record Vote No. 199). The companion S.J.Res.192 remains on the Senate calendar (Calendar No. 431) with no floor vote. That resolves the question left open in the July 13 note below: the disapproval path failed on the floor, and 5Q’s planning assumption, that WISeR remains operationally active through FY2027, stands. Confidence: HIGH.
Sources: U.S. Senate Press Gallery floor log, July 16, 2026; congress.gov and govinfo bill status for S.J.Res.192 and S.J.Res.198.
Current status — as of July 13, 2026
Congressional Review Act. GAO concluded on May 12, 2026 (B-337994) that the WISeR model notice is a rule that CMS failed to submit to Congress, opening an expedited-consideration window. Joint resolutions of disapproval were introduced May 19–20 (Sen. Wyden; Reps. DelBene and Landsman). As of this date, no Senate floor vote on S.J.Res.192 appears in the public record, and the fast-track window flagged by analysts (~July 11) has passed, leaving ordinary procedural paths, and the appropriations process, as the live vehicles. Even with passage, a presidential veto of the administration’s own model remains likely; 5Q’s planning assumption is that WISeR remains operationally active through FY2027.
Appropriations. The House Appropriations Committee adopted an amendment barring federal funds for WISeR, which suggests bipartisan concern, though it would not bind before October 1, 2026 and faces Senate negotiation.
Gold-carding. The exemption program was scheduled to launch July 6, 2026 in Washington State. Providers achieving a 90% affirmation rate on at least 10 requests during the assessment period are exempted from review, with rollout to additional states to follow. Post-launch operational reporting is not yet available.
Transparency litigation. The Electronic Frontier Foundation’s FOIA action against CMS, seeking disclosure on the AI tools used by model participants, remains the confirmed litigation front.
Patient impact. Press reporting from the six states documents delays and error patterns, including approximately one hundred patients queued for epidural procedures at one Washington academic health system, payment delays stretching multiples beyond the promised turnaround, and initial approval rates in Texas near 62% that rose to 84% after physician review.
Confidence map (v1.1, status-adjusted)
| Finding / development | Confidence | Basis / what would change it |
|---|---|---|
| CMS ordered Virtix to submit a CAP for 72-hour turnaround violations across five audit domains. | HIGH | Press reporting of CMS statements; congressional-office confirmation. Direct CMS letter not yet public. |
| Virtix disputes formal receipt of the CAP. | HIGH | On-record company statement to trade press. |
| WISeR notice is a CRA “rule” improperly withheld from Congress. | HIGH | GAO B-337994, read directly. |
| Contingent-fee model: participants compensated via a share of averted expenditures. | HIGH | CRS IF13133; CMS FAQ; concurring press reporting. |
| No Senate floor vote on S.J.Res.192 occurred within the expedited window. | MED | Absence-of-evidence finding across congressional tracking and trade press through July 13; a recorded vote would change it immediately. |
| Repeal outcome: veto likely; model operationally active through FY2027. | MED | Analyst projection; inherently predictive. |
| All six regional participants have failed CMS standards to comparable degrees. | LOW | Advocacy-source claim; only the Washington enforcement action is confirmed. |
| Applicability of Azar v. Allina notice-and-comment reasoning to WISeR challenges. | FLAG | Analogical application; no filed complaint verified as advancing it. Verify pincite and posture before external citation. |
Why this is the case study
WISeR is one of the first public examples of Medicare moving from published, rule-based adjudication to vendor-operated, AI-assisted review. Traditional Medicare adjudication had published logic and appeal pathways; the model arrived with neither published model documentation nor demonstrated operational monitoring. Six months later, the corrections are being made in public, under federal supervision, at the contractor’s expense. For every payer and health system deploying algorithmic utilization review, the lesson is that the oversight structure needs to be in place before deployment, because problems of this kind tend to surface within the first months of operation.
v1.0 · June 2026 · Initial six-month analysis
v1.1 · July 2, 2026 · Enforcement supplement: Virtix CAP, CRA posture, appropriations, gold-carding, revised confidence map
This page · July 13, 2026 · Status note updated (CRA window lapse; gold-carding launch date passed); confidence map status-adjusted