The Facts Behind Accusations That UnitedHealthcare Uses Error-Prone AI to Deny Claims

After the fatal shooting of UnitedHealthcare CEO Brian Thompson on Dec. 4, 2024, a rumor circulated online claiming the health insurance company uses an artificial intelligence (AI) tool with a 90% error rate to deny insurance claims from customers.

Social media users spread the claim on X (archive) and Reddit (archive), where one post received more than 79,000 upvotes at the time of this writing: 

It was never about helping people
byu/CorleoneBaloney inMurderedByWords

As some social media posts accurately noted, the claim originated from an allegation in a lawsuit against UnitedHealthcare and its parent company, UnitedHealth Group. The lawsuit also names NaviHealth, a technology-driven division of UnitedHealth, as a defendant. NaviHealth (now called Home & Community Care) developed the AI tool at the center of the lawsuit, which is called nH Predict.

It's true that UnitedHealth once deployed AI software to evaluate claims, though the extent to which the technology informed decisions about denying coverage, if any, is unknown as of this writing. It's also unclear whether the software is still part of the company's processes. UnitedHealth Group and UnitedHealthcare did not respond to Snopes' requests for information, and the lawsuit — which was filed on Nov. 14, 2023, in a federal court in Minnesota — is ongoing as of the publication of this story.

The California-based Clarkson Law Firm filed the suit on behalf of the families of two deceased patients who had UnitedHealthcare coverage. The families' attorneys allege UnitedHealth used an AI model that the company knew had a 90% error rate to deny care to patients with Medicare Advantage Plans, among other claims.

Medicare Advantage Plans, also known as "Part C," provide health coverage through private companies that participate in Medicare, the U.S. government's health insurance program for seniors and people with disabilities. 

Here's the claim in the lawsuit:  

This putative class action arises from Defendants' illegal deployment of artificial intelligence (AI) in place of real medical professionals to wrongfully deny elderly patients care owed to them under Medicare Advantage Plans by overriding their treating physicians' determinations as to medically necessary care based on an AI model that Defendants know has a 90% error rate. 

The complaint says the alleged error rate was based on the percentage of patient claims that the company initially denies and then gets "reversed through either an internal appeal process or through federal Administrative Law Judge (ALJ) proceedings." (The latter are formal hearings in which federal judges act as an impartial decision-maker.)

In a November 2023 interview with CBS News, a NaviHealth spokesperson, Aaron Albright, told the news outlet that the AI tool wasn't used to deny coverage but rather as "a guide to help [UnitedHealth] inform providers ... about what sort of assistance and care the patient may need." Albright said the lawsuit "had no merit." 

UnitedHealth argued in court that the lawsuit should be dismissed because the patients at the center of it did not complete appeals to their insurance claims being denied. 

While the court weighs the allegation, external investigations have uncovered evidence to indicate that nH Predict indeed increased insurance denial rates.

A March 2023 investigation by STAT News, a health care-focused publication, found that AI, in general, drove denials to "new heights in Medicare Advantage" and that one of the "biggest and most controversial companies" behind the emerging technology was NaviHealth. STAT News used federal records, court filings, confidential corporate documents and interviews with various people involved in the issue to produce the investigation, which was a Pulitzer Prize finalist in 2024. 

Months later, an Oct. 17, 2024, Senate investigation into denial rates for coverage via Medicare Advantage also found that UnitedHealthcare's rates "accelerated significantly once naviHealth began managing post-acute care," particularly when it came to coverage for patients in skilled nursing facilities. 

Data obtained by [the Senate's Permanent Subcommittee on Investigations] show that, while UnitedHealthcare's prior authorization denial rates increased for each type of post-acute facility during the period covered by this report, the increases were particularly striking for skilled nursing facilities. In 2019, the insurer issued an initial denial to 1.4 percent of requests for admission to a skilled nursing facility. But in 2022—the first full year in which naviHealth was managing them for UnitedHealthcare—the insurer denied 12.6 percent of such requests: in other words, its 2022 denial rate for skilled nursing facilities was nine times higher than it was three years before.

However, until there is conclusive evidence to prove that UnitedHealth currently uses AI technology to deny claims and that software has a 90% error rate, we consider the claim unproven.