AI is now denying health insurance claims with 90% error rate
FLORIDA, UNITED STATES — A federal lawsuit and a mounting record of reversed decisions are forcing scrutiny of how health insurers use artificial intelligence (AI) to deny coverage — starting with a UnitedHealth Group subsidiary whose AI-generated denial decisions were overturned on appeal nine times out of 10.
According to a report from WLRN, the figures come from naviHealth, a post-acute care AI platform UnitedHealth acquired for $2.5 billion in 2020 and deployed to automate coverage decisions for Medicare Advantage beneficiaries.
UnitedHealth AI denials face court scrutiny
naviHealth’s algorithm was designed to predict when patients no longer needed post-acute care. When patients appealed, nine in 10 predictions were reversed, according to the lawsuit.
“It used to be that before an insurer denied a claim, a human being had to look at [the claim]. These days, things have become much more automated,” said Jude Odu, founder of Health Cost IQ and a former UnitedHealth employee who published a book on AI-powered health plans in May 2026.
UnitedHealth Group is now facing a federal lawsuit over AI-generated denial documentation, with a judge setting an April 2026 compliance deadline — placing the industry’s reliance on automated denial systems under direct legal scrutiny.
AI amplifies disparities and provider burden
A separate AI scheduling system produced wait times 33% longer for Black patients — evidence that tools trained on historical data replicate and entrench existing disparities.
Odu said the mechanism is not incidental: “AI essentially takes the existing frameworks of discrimination and just magnifies them.”
CMS is already deploying AI-assisted coverage review at scale. Expansion into Medicaid and traditional Medicare decisions is expected to deepen the problem.
For providers, automated denials without human review translate directly into appeals volume, delayed reimbursement, and a compliance exposure that grows with every algorithmic decision that goes unchallenged.
As automated denial rates rise, providers face a growing secondary workload: documenting appeals, resubmitting claims, and managing prior authorization cycles that increasingly require overturning algorithmic decisions.
Revenue cycle management (RCM), prior authorization support, and denial appeals management are among the most commonly outsourced administrative functions in U.S. healthcare — a multibillion-dollar sector where specialized business process outsourcing (BPO) teams are absorbing the documentation and appeals workload that AI denial systems are adding to provider back offices.
For health systems and physician practices handling rising denial volumes, outsourcing appeals processing and prior authorization work to trained offshore teams is an operational response that doesn’t wait for regulators or courts.
