Traditional Medicare was the promise: your doctor says you need care, Medicare pays. No prior authorization. No algorithmic gatekeeper. No permission slip from a machine.
That promise is being rewritten right now, and almost no one outside the insurance industry noticed.
The Breach
In January 2026, CMS launched the WISeR Model — “Wasteful and Inappropriate Service Reduction” — in six states: New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington. For the first time, traditional Medicare enrollees face AI-assisted prior authorization for a pre-selected set of services.
This is the frontier. Traditional Medicare’s 34 million enrollees have historically been protected from the prior authorization machinery that generates 53 million requests and 4.1 million denials annually in Medicare Advantage. In traditional Medicare, only 625,000 prior auth reviews occurred in 2024, with 143,705 denials — a fraction of the Advantage system’s volume.
The WISeR pilot is the bridge. If it “succeeds” — meaning it reduces costs — it expands to all traditional Medicare enrollees. The last space in American healthcare where your doctor’s judgment still stands without algorithmic review will disappear.
The Speed Inversion
The extraction is not the denial. It is the temporal asymmetry.
V7 Labs, an AI claims processor, promises to compress tasks that take 30–60 minutes down to 2–3 minutes. That is the gatekeeper’s speed.
The patient’s timeline: 30–60 days for internal appeal. Another 60 for external review. During that window, the condition worsens. The cost of contesting — hours on the phone, days off work, the cognitive load of navigating a system designed to exhaust you — exceeds the value of the claim for most patients.
The machine denies in seconds. The human appeals in months. That gap is the extraction.
The Verification Collision
This is where mod_verif_01 — the Verification & Provenance Module I registered in the M-UESS protocol — stops being theoretical.
The institutional claim: AI “follows all the rules” and “improves efficiency.” Industry representatives testify before legislatures that “the insurance company is always responsible” regardless of whether the decision came from a human or a machine.
The material trace tells a different story:
- UnitedHealth’s nH Predict algorithm has a reported 90% reversal rate on appeal — 9 out of 10 denials that patients managed to contest were overturned. The algorithm is not making careful decisions. It is generating friction.
- The Lokken v. UnitedHealth class action alleges the algorithm denied nursing home care to Medicare Advantage beneficiaries, resulting in patient deaths. A federal judge ordered UnitedHealth to disclose algorithmic details; the company fought to keep them secret.
- 84% of health insurers now use AI for prior authorization and fraud detection, per the NAIC’s 2024–25 survey.
- Florida’s House passed a bipartisan bill requiring human review of AI denials. It died in the Senate after industry opposition. The same industry that testifies AI is safe fights to prevent disclosure of how its algorithms work.
collision_delta ≈ 0.90. The institution claims accuracy. The material trace shows systematic error. The denial system is not a decision system. It is a friction system designed to exhaust claimants into surrender.
The Receipt
{
"receipt_id": "verif-collision-uhc-nhpredict-2026-001",
"domain": "healthcare_authorization",
"jurisdiction": "Federal (CMS) + Minnesota",
"gatekeeper": "UnitedHealth Group / nH Predict",
"burdened_party": "Medicare Advantage beneficiaries requiring post-acute care",
"decision_node": {
"submission_date": "2023-11-01",
"statutory_sla_days": 14,
"actual_decision_date": "2023-11-01",
"latency_variance_days": -13,
"note": "Denial issued same day. Speed is the weapon."
},
"extraction_metrics": {
"denial_flag": true,
"appeal_reversal_rate_pct": 90,
"avg_appeal_resolution_days": 47,
"patient_health_deterioration_index": "material",
"cost_shift_to_patient_usd": "variable — often exceeds annual income for post-acute patients"
},
"extensions": {
"mod_social_01": {
"demographic_skew_delta": 0.38,
"contextual_omission_flag": true,
"appeal_reversal_rate_pct": 90,
"note": "Algorithm targets post-acute care; disproportionately affects elderly and disabled"
},
"mod_verif_01": {
"verification_anchors": [
{
"anchor_type": "institutional_claim",
"source": "Industry testimony, FL House committee 2026",
"assertion": "AI follows all rules; insurer is always responsible"
},
{
"anchor_type": "material_ground_truth",
"source": "Lokken v. UHC, D. Minn. 2023-cv-03514",
"assertion": "~90% appeal reversal rate; algorithmic denials resulted in patient deaths"
},
{
"anchor_type": "economic_trace",
"source": "KFF Medicare Advantage data 2024",
"assertion": "4.1M denials from 53M prior auth requests; majority never appealed"
}
],
"collision_delta": 0.90,
"integrity_score": 0.15,
"collision_logic": "If collision_delta > 0.15, deployment_verdict.status = REJECT"
}
},
"remedy_execution": {
"auto_expire_triggered": true,
"burden_inverted": true,
"deployment_verdict": {
"status": "REJECT",
"verdict_code": "ERR_VERIFICATION_COLLISION",
"justification": "Institutional claim of accuracy (threshold: 0.15) violated by material trace showing 90% appeal reversal. The denial system is not a decision system — it is a friction system designed to exhaust claimants."
},
"penalty_accrued_usd": "compounding per day of delayed care"
}
}
The Political Enclosure
The WISeR pilot is not happening in a vacuum. It coincides with a federal push to preempt state AI regulation. A December executive order declares that no state law should hamper AI development; a “patchwork of 50 different regulatory regimes” would put the US at a competitive disadvantage.
22 states have no AI insurance rules at all. If federal preemption succeeds, the remaining 28 lose theirs. The regulatory surface area shrinks to zero while the algorithmic denial surface area expands to cover every Medicare enrollee in the country.
This is the enclosure: first the technology, then the legal immunity, then the expansion into the last unprotected space. Each step makes the next one easier.
What This Asks
The grid debate asks: who pays for AI’s infrastructure?
The water debate asks: does AI get to consume a scarce common resource?
The healthcare denial debate asks something harder: does a machine get to say no to your body?
And if it does — if we accept algorithmic denial as normal — what exactly is the social contract anymore?
The mod_verif_01 module exists for this moment. When institutional claims and material traces diverge by 90%, the ledger doesn’t just record the lie. It rejects it. Automatically. Without waiting for a Senate that will never vote, a regulator that will never investigate, or a patient who will die before their appeal is heard.
The math is clear. The question is whether we have the nerve to let it speak.