Images chosen by Narwhal Cronkite
Artificial intelligence tools are playing an increasingly prominent role in how American health insurers evaluate and deny coverage claims — a practice drawing lawsuits, state legislation, and growing concern from patients and medical professionals.
How AI Is Being Used to Process Claims
Health insurers have adopted AI systems to automate the prior authorization process — the gatekeeping step that determines whether a patient’s proposed treatment will be covered before it takes place. According to recent industry data, approximately 84 percent of health insurers now use some form of AI in prior authorization decisions. These tools can process claims in a fraction of a second, scanning diagnostic codes, medical records, and treatment histories against preset criteria.
Insurers say the technology improves efficiency and reduces administrative costs. A UnitedHealth spokesperson pointed to the company’s AI Review Board — a team described as including clinicians, scientists, and other experts — as evidence of ongoing oversight. Industry group America’s Health Insurance Plans has characterized responsible AI use as a way to make claims management faster and more effective for both patients and providers.
But a growing body of lawsuits, investigative reports, and patient accounts suggests the systems are also producing a surge in coverage denials, sometimes without any meaningful human review of the individual case.
The Lawsuits Piling Up
Two high-profile class-action lawsuits have put the issue in sharp focus. UnitedHealth Group faces a suit alleging that an AI tool called nH Predict — developed by its subsidiary NaviHealth — was used to deny extended care to elderly Medicare Advantage patients based on algorithmic outputs rather than individual medical review, in some cases contributing to patient deaths. The company denies the claims and says the tool is intended as a clinical guide, not a decision-maker.
Separately, Cigna faced a lawsuit alleging its PXDX algorithm allowed physicians to automatically deny thousands of claims at a time for treatments that didn’t match preset codes — without reviewing individual patient records. An investigation found that reviewers were spending an average of roughly 1.2 seconds per case during some periods. Cigna has disputed the characterization of its review process.
A court in Minnesota recently allowed discovery into UnitedHealth’s use of the nH Predict system, ruling that patients are entitled to understand how the AI works, what its development goals were, and whether it was designed to override physician judgment. The decision is seen as a significant step toward broader judicial scrutiny of AI use in insurance.
Data from the Regulatory Review indicates that approximately 90 percent of UnitedHealth’s denied claims that were appealed were ultimately overturned by federal administrative law judges — a figure that critics argue illustrates a systemic problem with the initial denial process. Yet fewer than one percent of patients ever file an appeal.
States Push for Regulation
In response, lawmakers in more than a dozen states are considering or have introduced legislation to limit how AI can be used in claims decisions. Minnesota is considering a bill that would prohibit insurers from using AI as the sole basis for denying a prior authorization request. Florida attempted similar legislation in 2025 requiring human review for AI-generated denials, but the bill passed the House before dying in the Senate. At the federal level, the Centers for Medicare and Medicaid Services issued guidance requiring that coverage determinations account for the patient’s individual circumstances — though critics argue the rule falls short of meaningful enforcement.
Initial claim denial rates hit 15 percent in 2026, according to industry tracking data, with AI-triggered rejections up roughly 9 percent since 2022. The American Medical Association has reported that 61 percent of physicians say AI use by health plans is increasing prior authorization denials and creating unnecessary barriers to patient care.
What Patients Can Do
Healthcare advocates emphasize that denied claims are far from final. Studies suggest that 40 to 90 percent of appeals succeed when patients formally contest a denial — yet the process is often lengthy, confusing, and exhausting, particularly for those who are already seriously ill.
Patients who receive a denial are entitled to a written explanation and have the right to appeal through their insurer and, if unsuccessful, through external review. Including a doctor’s letter, relevant test results, and peer-reviewed research supporting the treatment in an appeal significantly improves outcomes, according to patient advocacy groups.
The broader question — of how much weight an algorithm should carry in decisions about medical necessity — remains unresolved, both legally and politically. As courts, state legislatures, and federal regulators continue to weigh in, the outcome will shape how tens of millions of Americans interact with their health insurance for years to come.
