Trump Administration Plans AI-Powered Medicare Denials

The Trump Administration is moving to significantly expand prior authorizations (PAs) within Traditional Medicare, a mandatory pre-approval process for healthcare services to be covered by insurance. This initiative involves a public-private partnership model that plans to leverage artificial intelligence (AI) with concerning incentives for participating companies.

On June 27, 2025, the Centers for Medicare and Medicaid Services (CMS) Innovation Center unveiled the Wasteful and Inappropriate Service Reduction (WISeR) Model, slated to run from 2026 to 2031. The stated objective of WISeR is to curb wasteful spending in Medicare, citing a 2022 Medicare Payment Advisory Commission (MedPAC) report that found $5.8 billion in “low-value care” – services with minimal or no clinical benefit, or where risks outweigh benefits.

Under the WISeR model, CMS intends to contract with “companies with expertise providing recommendations on medical necessity of coverage for payers using enhanced technology like AI.” While these companies will utilize AI, CMS specifies that human clinicians must ultimately validate medical determinations regarding whether a service is deemed wasteful. Currently, Traditional Medicare has a limited number of services requiring prior authorization; however, WISeR aims to substantially broaden this scope, drawing on the prior authorization experience of companies active in Medicare Advantage.

Concerns Over Prior Authorizations and Patient Care

Despite its stated goal, the introduction of widespread prior authorizations, particularly within this public-private framework, is raising alarms about potential harm to patients. Prior authorizations are already prevalent in privatized Medicare Advantage and standard private insurance, where they frequently lead to delays and denials of vital medical coverage.

A 2024 survey by the American Medical Association (AMA) of 1,000 physicians revealed that a staggering 93 percent reported PAs causing delays in necessary care (15% always, 42% often, 36% sometimes). Furthermore, 82 percent of physicians observed patients abandoning recommended care due to difficulties with the PA process (2% always, 20% often, 60% sometimes).

In 2023, approximately 50 million PA requests were filed for Medicare Advantage patients, with insurance companies denying around 3.2 million (6.4%). Although patients and providers appealed only 11.7% of denials, they boasted an impressive 81.7% success rate. A 2022 Government Accountability Office (GAO) report examining a random sample of MA PA denials found that 13% were improper, meaning the care should have been covered under Medicare rules. If this trend continued into 2023, it suggests that PAs could have resulted in approximately 112,000 improperly denied treatments, as only a small fraction of denials are successfully appealed.

Increased Administrative Burden and Costs

The proliferation of prior authorizations also imposes a significant administrative burden on healthcare providers, likely escalating overall costs. The AMA survey indicated that physicians spend an additional 13 hours weekly on PA tasks, with 40 percent of physicians dedicating staff solely to handling PAs.

The U.S. healthcare system already grapples with higher administrative costs compared to other developed nations, contributing to worse health outcomes despite double the per-person spending. Much of this stems from the complexities of the predominantly for-profit, private insurance model. In contrast, Traditional Medicare has historically maintained much lower administrative costs, with only 1.1 percent of spending allocated to administration in 2023, far below the 12-18 percent seen in private insurance over the last decade. Expanding PAs in Traditional Medicare risks increasing not only the administrative load on providers but also direct taxpayer costs through higher administrative spending.

The Peril of AI in Denying Care

A critical concern with the WISeR model is its potential to foster corporate exploitation within Traditional Medicare through its contracting mechanisms. A detailed October 2024 report by the US Senate Permanent Subcommittee on Investigations highlighted how three major Medicare Advantage insurers – United Healthcare, Humana, and CVS – disproportionately denied PA requests for post-acute care.

Internal documents from United Healthcare even promoted their Machine-Assisted Prior Authorization (MAP) technology, while CVS began deploying an AI program in 2021 specifically to reduce post-acute care approvals. These systems, as documented by a March 2022 CVS meeting, were designed to increase denials, with notes indicating that reducing PAs would lead to “financial losses too large to move forward.”

The WISeR model structure offers a “percentage of the savings associated with averted wasteful, inappropriate care” as a reward to contracted companies using AI for reviews. This incentivizes these third-party, for-profit entities to deny as much care as possible to maximize their financial gains, raising serious doubts about their commitment to prioritizing patient access to necessary healthcare over profit margins.

Beyond these perverse financial incentives, the WISeR model reflects the Trump administration’s continued, often unsubstantiated, reliance on AI. The Senate report elaborated that the use of AI and machine models is intended to boost denial volumes. The Medicaid and CHIP Payment and Access Commission (MACPAC) has also warned that companies can bias and manipulate algorithms to unfairly emphasize care denials.

The reliability of AI for determining healthcare coverage for American seniors remains unproven. Reports on the FDA’s “Elsa” AI model indicate its struggles with factual accuracy, and the White House’s “Make America Healthy Again” (MAHA) Report was plagued by incorrect, AI-generated references to non-existent studies.

A Question of Priorities: Addressing True Waste

While the government’s interest in preventing $5.8 billion in unnecessary and inappropriate care is legitimate, the proposed WISeR approach raises significant red flags. The combination of: (1) introducing PAs that have a documented history of delaying and denying vital services in private insurance, (2) adopting a CMS model that increases administrative costs in an already expensive healthcare system, (3) granting perverse incentives to for-profit companies to deny care for profit maximization, and (4) relying on unproven AI models for life-saving care decisions, suggests this approach could ultimately harm Americans more than it helps.

Crucially, the WISeR model must be viewed within the broader context of healthcare policy. MedPAC estimated that CMS will waste an astounding $84 billion in 2025 by overpaying private insurance companies in Medicare Advantage (such as United Healthcare, Humana, Aetna, and Elevance Health). Other recent estimates have placed annual overpayments even higher, ranging from $80 billion to $140 billion. Based on MedPAC’s 2025 projection, the Committee for a Responsible Federal Budget forecasts a staggering $1.2 trillion in MA overpayments over the next decade.

Despite suggestions from some Republican senators to include provisions in recent reconciliation legislation targeting “upcoding” – a key method insurers use to inflate patient sickness levels for higher taxpayer reimbursements – the Trump Administration and congressional Republicans ultimately did not push for any measures addressing this massive source of waste, which is more than 14 times larger than the $5.8 billion WISeR aims to tackle.

While curbing wasteful spending in American healthcare is a valid governmental objective, the answer is not funneling money to for-profit companies by incentivizing them to deny healthcare coverage through unproven, manipulable AI models. Far more impactful waste, stemming from insurer overpayments, hospital pricing, and pharmaceutical practices, remains largely unaddressed.