On March 25, 2026, the Electronic Frontier Foundation filed a federal lawsuit against CMS over a six-state pilot program called WISeR, the Wasteful and Inappropriate Service Reduction Model. The pilot uses AI to make prior authorization determinations in traditional Medicare. The lawsuit raises serious questions about transparency, accountability, and the quality of those determinations.

This is the first federal court challenge to CMS’s use of AI in prior authorization. For practices operating in the six pilot states (Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington), and for the broader question of where AI sits in adjudication, the case is worth understanding in some detail.

What WISeR Is

WISeR launched January 1, 2026 as a six-year demonstration pilot through CMS’s Innovation Center, running through 2031. Under the pilot, providers in the six participating states submitting traditional Medicare claims for a defined set of services must obtain prior authorization through an AI-driven review run by contracted private technology vendors.

CMS’s stated goal is to reduce ‘low-value’ care that has limited clinical benefit. Prior authorization has historically been rare in traditional fee-for-service Medicare. WISeR is essentially a structured test of whether AI-driven PA can be used as a cost-control mechanism in original Medicare in a way it already is in Medicare Advantage.

The Numbers That Triggered the Lawsuit

Reporting from the Washington Post, cited in the EFF complaint, surfaced approval-rate data from Texas that is hard to read as anything other than a structural problem:

• 62% of WISeR prior authorization requests in Texas were initially approved by the AI system
• 84% were ultimately approved after human review
• 92% are approved nationally in Medicare Advantage prior authorization

The gap between 62% (initial AI decision) and 84% (after human review) is the same pattern documented in private MA prior authorization. AI-driven first-pass denials get reversed at a high rate when an actual clinician looks at them. The new wrinkle is that this pattern is now operating inside traditional Medicare, on a population that historically has not faced PA at this scale.

What the Lawsuit Actually Argues

EFF filed under the Freedom of Information Act. The suit asks the court to compel CMS to release records that CMS has not produced in response to an earlier FOIA request, including:

• Agreements with the software vendors participating in WISeR
• Records related to tests for accuracy, bias, or hallucinations in the vendors’ AI technology
• Records related to any audits, monitoring, or evaluation of WISeR and participating vendors

The lawsuit is not (yet) a substantive challenge to whether CMS can use AI in PA. It is a transparency action. But the implicit argument is that the public, and the providers operating under WISeR, do not currently have access to the information needed to know whether the AI is being run responsibly. The complaint cites concerns about algorithmic bias, the quality of training data, and the financial incentives baked into vendor contracts that may reward denials.

What This Means for Practices in the Six States

If your practice operates in Arizona, New Jersey, Ohio, Oklahoma, Texas, or Washington, and you treat traditional Medicare patients in the service categories WISeR covers, a few things are worth doing now:

• Track your WISeR denial rate carefully. If your first-pass approval rate is materially below the 62% Texas figure, document it. That kind of data may matter for future advocacy and is independently useful for your operations.
• Build a default-to-appeal workflow for WISeR denials. The 22-percentage-point gap between AI-decision approval and post-review approval suggests appeals are highly likely to succeed. Standard payer-tier appeal logic applies.
• Use the new specificity requirements. CMS-0057-F requires specific reasons for PA denials. WISeR denials are not exempt. If a denial reason is vague, that is an appeal lever.
• Maintain detailed clinical documentation. Where AI-driven adjudication is reviewing your PA requests, the quality and specificity of the clinical documentation you submit upfront is the single biggest factor in whether the request clears on first pass.

The Broader Question

Beyond the specific six-state pilot, WISeR is a test case for whether the federal government can use AI-driven PA as a long-term cost-control mechanism across Medicare. The outcome of EFF’s lawsuit, and CMS’s eventual evaluation of WISeR, will shape that decision for years.

In the meantime, the practical message for practices is unchanged. When AI is in the adjudication path, the gap between first-pass denials and post-review approvals is consistently wide. Treating that gap as appealable revenue, rather than absorbing it as denial volume, is the workflow change that pays.

How HSC Approaches AI in the Adjudication Path

Harris Secure Connect’s role is to make sure the foundation under any AI-driven workflow is solid. Clean clinical documentation submission, accurate eligibility data, structured ERA reconciliation, and the audit trail your team needs to reconstruct what happened when something goes sideways. As AI enters more parts of the adjudication path, the infrastructure under it matters more, not less.

Related Resources

• EFF press release on WISeR lawsuit
• Healthcare Dive coverage of the EFF v CMS lawsuit
• CMS WISeR Model overview

If you operate in a WISeR pilot state and want to think through how to handle AI-driven denials at scale, our team is happy to walk through what stronger documentation and appeal workflows look like.