Trump administration using AI to deny Medicare claims for seniors

By Splitfeed EditorialLast updated Sep 26, 2026, 19:06 UTC

On September 25, 2026, reporting from Ars Technica described the Trump administration deploying artificial intelligence systems to adjudicate Medicare claims for seniors. The reporting characterizes the program as an experiment in AI-driven claim adjudication and raises a specific structural concern: that the vendors operating these AI systems have a financial incentive to deny as many claims as possible. No administration officials, specific vendor names, contract values, or denial rate figures were identified in the available sourcing.

The story matters because Medicare is the primary health coverage for tens of millions of Americans over 65, and the decision to approve or deny a claim can determine whether a senior receives surgery, medication, or long-term care. Introducing automated systems into that process raises questions that go beyond efficiency. If vendors profit by minimizing approvals, the financial incentive is structurally misaligned with the coverage guarantee Medicare is supposed to provide. What remains genuinely in dispute is whether this program has produced wrongful denials at scale, whether human review is part of the process, and what legal or regulatory oversight governs how the AI reaches its decisions.

With no published reactions yet from the Pro-AI, Anti-AI, or Middle Ground camps, their likely positions can be sketched from the broader patterns each camp typically holds. Pro-AI voices would be expected to argue that algorithmic adjudication reduces fraud, speeds up legitimate claims, and can be more consistent than overburdened human reviewers, though the vendor incentive structure would likely be an uncomfortable detail for that argument. Anti-AI voices would be expected to treat this as confirmation of a recurring concern, that deploying AI in high-stakes welfare systems without adequate transparency or accountability causes direct harm to vulnerable people. Middle Ground observers would be expected to draw a distinction between automation itself and the specific contractual design, arguing that the problem here is the incentive structure rather than AI use in principle.

The argument is likely to sharpen once specific denial rate data, vendor contracts, or congressional or inspector general scrutiny becomes public. Any administrative or legal challenge that forces disclosure of how the AI is trained, what criteria it applies, and how often its decisions are reviewed by humans would materially change what can be claimed by any side.