82% of Americans back a law requiring human review of AI claim denials, survey finds

Eighty-two percent of Americans would support a law requiring a human to review any medical claim denied by an artificial intelligence (AI) tool, according to an open enrollment survey eHealth released Oct. 28.

Seven states passed laws this year limiting how health insurers use AI in coverage decisions, and a bipartisan House bill introduced Sept. 1 would require a licensed health care professional to make any private insurance denial that turns on clinical judgment.

eHealth, an online health insurance marketplace and licensed agency that says it uses AI to help consumers compare plans, surveyed more than 1,000 U.S. adults Oct. 7-9 through SurveyMonkey. The margin of error is plus or minus 3 percentage points. The report does not say whether any respondents had a claim denied by AI.

Where the law stands

Laws in Washington, Iowa and Alabama took effect this year. Colorado, Georgia and Utah follow Jan. 1, 2027, and an Illinois law requiring a person to make or review every downcoding decision takes effect Jan. 1, 2028. They join earlier statutes in California, Arizona, Texas, Nebraska, Connecticut and Maryland, according to a tally by RISE.

The state laws don't reach self-insured employer plans, which federal law governs, and Medicare Advantage plans generally follow federal standards.

Traditional Medicare began requiring prior authorization for a set list of services in Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington on Jan. 1 under the Centers for Medicare & Medicaid Services (CMS) Wasteful and Inappropriate Service Reduction (WISeR) Model.

Technology companies review the requests using AI and other tools and are paid a share of the spending they avert. A clinician employed by the company must review every request it declines to approve, according to a CMS fact sheet.

Two of those companies denied 5,944 prior authorization requests in the program's first three months, according to roughly 1,000 pages of CMS records the Electronic Frontier Foundation obtained through a Freedom of Information Act lawsuit and released in September. Virtix, the Washington vendor, denied 53% of the more than 6,000 decisions it made through the end of March, according to a status report it sent CMS.

A contractor in Arizona made 14,300 decisions over the same period and denied fewer than 20%. Virtix has said its approval rate has since risen to nearly 70%. CMS ordered Virtix to file a corrective action plan in June for missing required turnaround times.

Thirty-one House members led by Rep. Suzan DelBene (D-Washington) wrote to CMS Administrator Mehmet Oz, M.D., in June, citing "care denials, long wait times, and increased administrative burden" in the six states.

They asked for state-by-state data on denials, appeals and overturns, and for how often peer-to-peer reviews are conducted by a physician in the same or a similar specialty as the treating physician. CMS has not published a response to the letter. The agency has said it will release a six-month analysis of the model's outcomes, including turnaround times and preliminary accuracy data.

The Doctors Not AI Act of 2026 (H.R. 10210), introduced by Rep. Greg Landsman (D-Ohio), Rep. Buddy Carter (R-Georgia), Rep. Kim Schrier, M.D. (D-Washington) and Rep. Tom Barrett (R-Michigan), would bar an AI system from issuing or dictating any denial that involves clinical judgment.

A licensed professional with training and experience in the service at issue would have to make the decision after independently evaluating the patient's circumstances, and insurers would have to disclose when AI was used in claims review. The bill applies to group health plans and to insurers selling group or individual coverage, not Medicare or Medicaid. The Medical Group Management Association (MGMA) endorsed it Oct. 16.

"It is critical that doctors, not automated systems or AI-powered models, are in charge of critical healthcare decisions for their patients," Schrier said in a statement when the bill was introduced.

Denials that don't survive appeal

Medicare Advantage plans overturned 95% of appealed prior authorization denials for skilled nursing facility admission, the Department of Health and Human Services Office of Inspector General (OIG) found in a June report on 19 insurers' June 2024 decisions. Patients and their physicians appealed only 18% of those denials. The overturn rate "indicates that some enrollees were initially denied medically necessary care," the OIG said.

The contractor naviHealth, a UnitedHealth Group subsidiary, processed half of the requests, denied 14% of them and saw 97% of its appealed denials overturned, according to the OIG. naviHealth is behind nH Predict, the algorithm at the center of a class action lawsuit filed in Minnesota in 2023 by former UnitedHealthcare Medicare Advantage members or their heirs, who allege the company used the tool to cut off post-acute care.

The case is in discovery. UnitedHealth has called the allegations unfounded and says the tool is used only as a guide, with coverage decisions made by medical directors.

UnitedHealthcare's prior authorization denial rate for post-acute care rose from 10.9% in 2020 to 22.7% in 2022, according to a 2024 Senate Permanent Subcommittee on Investigations report that also documented the largest Medicare Advantage insurers' growing use of automation and predictive technology in those decisions.

Sixty-one percent of physicians said they were concerned that health plans' use of AI is increasing prior authorization denials in a 2024 survey by the American Medical Association (AMA) of 1,000 practicing physicians. "Emerging evidence shows that insurers use automated decision-making systems to create systematic batch denials with little or no human review, placing barriers between patients and necessary medical care," then-AMA President Bruce A. Scott, M.D., said in a statement released with the survey.

"It's so overwhelming now, especially with payers using AI and things like that," Shawntea "Taya" Gordon, MBA, FACMPE, CEO of Atlas & Perpetua Healthcare Consulting, told Physicians Practice at MGMA's annual conference in San Antonio. "They're able to review things faster, deny things faster, and we're still pretty manual in terms of our appeals efforts."

"Artificial intelligence is never used to deny care. That's non-negotiable," Brian Covino, M.D., chief medical officer of Cohere Health, which sells prior authorization technology to health plans, told Medical Economics in August.

He said the company's tools have lowered denial rates by flagging missing information, such as a required imaging study, while a request is being submitted.

Patients are asking AI first

Forty-eight percent of respondents said they have used an AI tool to review their health insurance options, and 85% of those said they chose their current plan based in full or in part on advice from AI.

Among Americans who have used AI for medical advice, 67% said that advice has sometimes been incorrect. Twenty-two percent of all respondents said AI can already replace doctors for medical advice, 18% expect that within five years and another 18% in five to 10 years. Thirty percent said it will never happen.

In an eHealth survey released in June, 49% of insured Americans said they had used AI tools for medical advice, and 63% of those said they acted on the guidance without consulting a doctor.

Cost and coverage confusion

Fifty-eight percent of insured respondents said concerns about out-of-pocket costs led them to skip or delay medical care in the past year, and 46% said confusion about their coverage did.

Eighty-four percent agreed that "it feels like the health insurance system is designed to be confusing on purpose." Fifty-seven percent called open enrollment stressful, including 28% who called it very stressful, and 71% said they have stayed in the same plan for multiple years to avoid researching their options.

"Nobody sits you down in high school and tells you how to read your insurance card," Gordon said in the same interview. "They don't explain what a prefix is, what a prior auth is, what your co-insurance and deductible are."

Patients who understand their coverage, she said, better understand "how to pick the most appropriate plan for them next year when open enrollment happens" and why the front desk is asking them for money.

Among enrollees in Patient Protection and Affordable Care Act (ACA) plans, 68% said they worry they won't find affordable coverage during open enrollment, including 40% who strongly agreed.

Thirty-six percent called their current plan unaffordable, and 37% said their government premium subsidies were reduced for 2026 compared with 2025. Excluding enrollees who don't receive subsidies, that figure was 45%.

Enhanced ACA subsidies introduced during the COVID-19 pandemic lapsed Jan. 1, 2026, the report notes, though changes in income can also affect subsidy amounts.

Among people with employer-sponsored coverage, 77% wish their employer offered a broader selection of plans, though 79% said their current plan is a good match for their needs and budget.

Seventy-eight percent of likely voters said health care affordability is among their top three issues in the Nov. 3 midterm elections, including 28% who called it their top issue.

What's next

The Colorado, Georgia and Utah laws take effect Jan. 1, the same day Iowa begins requiring that denials and downgrades of medical necessity requests come from a reviewer in the same or a similar specialty as the requesting clinician.

Health plans also face a Jan. 1, 2027, deadline under CMS-0057-F, the CMS Interoperability and Prior Authorization Final Rule, to run electronic prior authorization systems. The Doctors Not AI Act has been referred to the House Energy and Commerce Committee, with additional referrals to the Ways and Means and Education and Workforce committees.