Medicare AI Pilot Faces Scrutiny After Reports of Delays and Denials

THE SIGNAL · September 25, 2026

A federal pilot that uses artificial intelligence in Medicare prior-authorization decisions is continuing despite reports of delayed care, technical problems, and disputed denials. The program, known as WISeR, raises a practical question larger than one Medicare experiment: what safeguards are needed when AI becomes part of a process that can affect whether and when a patient receives treatment?

What happened

According to Ars Technica, the administration began a WISeR pilot in January that uses AI to authorize or deny certain types of care for Medicare patients. Medicare is the federal health-insurance program for seniors, and the outlet reports that the affected care had not previously required this type of prior authorization.

Prior authorization generally requires approval before a service can proceed. In the WISeR pilot, AI is part of a decision process that can determine whether care is authorized or denied. Ars Technica reports that problems emerged soon after the launch, including technical difficulties, long waits for decisions and care, denials that providers found difficult to understand, and frustration among doctors and patients.

The outlet said that documents released earlier this month by the Electronic Frontier Foundation, obtained through litigation, included feedback from healthcare providers. One provider described the program as a disgrace and reported seeing patients in pain while awaiting care. Those documents, Ars Technica reported, broadly confirmed the earlier accounts of delays and problematic decisions.

Questions about the program are not limited to its operation. Ars Technica reports that the Government Accountability Office concluded in May that officials did not follow proper procedure in establishing the pilot, raising questions about its legal basis. Lawmakers have sought information about the program and have attempted to stop it, according to the report. Even so, the pilot appears to be continuing, with expansion planned in coming years.

Why it matters

AI systems are increasingly used to sort, summarize, recommend, and automate work. Healthcare makes the stakes of that trend unusually clear. A flawed recommendation in an administrative workflow may create extra work; a flawed or delayed authorization can leave a patient waiting for treatment. The central concern here is not simply whether an AI system can process claims quickly, but whether the full process produces timely, understandable, and accountable decisions.

The reported issues also show why describing a system as AI-assisted does not answer the most important questions. Patients and providers need to know what information is being evaluated, how an adverse decision can be challenged, who can override it, and how long appeals or reviews take. The packet does not establish which specific model or rules produced any particular denial. But the accounts reported by Ars Technica indicate that implementation problems can have immediate consequences even when the underlying tool is used within an administrative program.

There is also a governance issue. When an AI-linked program operates in a public-benefit system, procedural compliance, public documentation, and independent oversight matter alongside technical performance. The GAO finding cited by Ars Technica concerns how the pilot was set up, while the provider feedback concerns its effects in practice. Together, those are separate but related tests: whether the program was authorized and designed properly, and whether it is working safely for people who rely on it.

What to watch

  • Whether officials publish more detail on the AI system’s role in authorization and denial decisions.
  • Whether patients and providers receive clearer routes for review, appeal, and human intervention when care is delayed or denied.
  • How the administration responds to lawmakers’ questions and to the GAO’s procedural concerns.
  • Whether any planned expansion proceeds before the reported operational problems are addressed.

Source

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

Leave a Reply