Trump admin using AI to deny medical care for seniors in disastrous experiment

The Trump administration's WISeR pilot is using AI-driven prior authorization for certain Medicare care, triggering reports of technical failures, lengthy treatment delays, disputed denials and patient harm. Federal documents obtained by the Electronic Frontier Foundation corroborated provider complaints, while the GAO found in May that officials failed to follow proper implementation procedures, raising legal concerns. Despite congressional efforts to investigate or halt the program, WISeR is continuing and is planned to expand in coming years.

Analysis

The direct earnings impact is likely immaterial because the program sits in traditional Medicare rather than Medicare Advantage (MA), but the political spillover matters for MA insurers’ utilization-management economics. HUM is the highest-beta exposure given its concentrated MA earnings base and relatively thin margin cushion; UNH, CVS and CNC have more diversified earnings but materially larger absolute exposure to any mandated reduction in authorization denials or appeals-related administrative costs. A bipartisan backlash could turn a narrowly scoped pilot failure into a broader CMS rulemaking agenda on algorithmic transparency, human review, turnaround times and auditability.

Near term, this is more headline and congressional-oversight risk than a fundamental catalyst. Over the next 1-3 months, monitor whether CMS pauses expansion, releases denial/appeal data, or whether congressional inquiries explicitly connect the pilot to MA utilization practices; those events would increase the probability of 2027 pricing and medical-loss-ratio pressure. Over 6-18 months, a requirement for human-in-the-loop review would favor scaled platforms able to absorb compliance costs, while reducing the margin value of automated utilization management for smaller MA plans.

The consensus risk is not that AI itself is unusable, but that public-sector implementation failures create a regulatory precedent that insurers cannot dismiss as limited to fee-for-service Medicare. Conversely, a shutdown of the pilot without MA-specific action would likely make any insurer selloff a buying opportunity: traditional-Medicare authorization changes do not mechanically alter MA reimbursement or benefit economics. The thesis is falsified if CMS publishes evidence of low overturn rates, timely approvals and a legally durable expansion framework, reducing the likelihood of broader intervention.

Key Decisions for Investors

• Do not establish a directional AI-healthcare trade on this development alone; direct revenue sensitivity for listed issuers is unverified. Set an alert for CMS expansion, a formal GAO/legal remedy, or MA-focused congressional action.
• Maintain a cautious relative-value bias: short HUM versus long UNH over a 1-3 month oversight window if HUM materially outperforms, as HUM has greater MA-policy beta while UNH’s Optum and commercial businesses diversify the exposure. Exit if CMS limits action to the traditional-Medicare pilot or HUM reiterates MA margin guidance without elevated authorization costs.
• Avoid adding to MA-insurer longs into any CMS data release until denial rates, appeal overturn rates and required human-review standards are known. A broad rule affecting MA plans would be more consequential than the current pilot and could pressure 2027 bids before visible P&L impact.
• Watch health-services and workflow vendors for second-order opportunity rather than assuming AI beneficiaries win: vendors with auditable clinical workflow, appeals and documentation capabilities could gain compliance spend, while opaque denial-automation offerings face product and reputational risk. No ticker-specific recommendation is warranted without confirmed contract exposure.