Health Insurers Deploy AI Across Medical Reviews, Raising Questions About Denials and Oversight
Artificial intelligence now plays a role in nearly every decision a health insurer makes about whether to cover a medical procedure. Insurers use AI for prior authorization before treatment, concurrent review during care, and retrospective review after services are delivered. According to the National Association of Insurance Commissioners, 84% of health insurers reported using AI or machine learning across multiple product lines, with 68% specifically using it for prior authorization.
Insurers say the technology helps identify whether care is medically necessary. What's less clear to patients and providers is how much AI actually drives denial decisions, and whether humans adequately oversee those systems.
The Financial Incentive
Major health insurers facing tighter profit margins are accelerating AI adoption. During first-quarter earnings calls this year, executives at UnitedHealth Group told investors they planned to use AI to cut $1 billion in costs. Other large insurers made similar announcements.
Jennifer D. Oliva, a law professor at Indiana University Bloomington who studies how insurers use AI, said the technology has been embedded in insurance operations for 10 to 15 years. But the scope and speed of deployment has increased significantly.
The practice of having humans monitor AI decisions varies widely. Many states and health plans require human oversight, and some mandate that a medical professional be involved in reviews. Insurers often disclose these requirements in fine print at the end of documents, Oliva said.
Medicare's AI Experiment and Legal Challenge
The Centers for Medicare and Medicaid Services launched a five-year pilot program this year called the Wasteful and Inappropriate Service Reduction (WISeR) Model. It operates in six states-Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington-to test whether AI can streamline prior authorization for Medicare services.
Providers in participating states must obtain AI-backed prior authorization before delivering certain services to traditional Medicare patients. Early reports show problems. Providers have reported denials for care that fell within coverage guidelines. Some patients waited longer than usual for pain relief and minimally invasive procedures for spinal compression fractures.
On March 25, the Electronic Frontier Foundation filed a Freedom of Information Act lawsuit against CMS seeking details about how WISeR algorithms work. The organization raised concerns about algorithmic bias, privacy violations, and wrongful denials. It also noted that vendors administering the program in each state share in cost savings tied to reduced utilization-creating a financial incentive to deny care.
State-Level Regulation Emerging
With little federal oversight of AI use in insurance, individual states are moving ahead with legislation. California, Colorado, Illinois, and Utah have considered bills to limit health insurers' use of AI. In February, the Alabama Senate passed legislation requiring a physician or other qualified provider to be involved in AI-related prior authorization denials, with a potential effective date of October 1. A Florida bill requiring human review of all care denials passed the House but stalled in the Senate.
Faster Decisions, More AI
On April 10, CMS issued a proposed rule requiring payers to support electronic prior authorization for some medications and make decisions within shorter timeframes. Under the rule, certain insurers would have 24 hours to approve or deny prior authorization requests for covered outpatient drugs, with an effective date as soon as October 1, 2027.
The compressed timeline raises a practical question: whether insurers will rely more heavily on AI to meet tighter deadlines, with less human review.
What to Watch
The WISeR program and related litigation will likely produce records about how insurers train, monitor, and audit AI systems. State legislation continues to move through legislatures. And the CMS prior authorization rule could reshape how quickly insurers make coverage decisions.
For healthcare professionals, understanding how AI affects coverage decisions-and where human oversight falls short-matters directly to patient care.
Learn more: AI for Healthcare and AI for Insurance cover developments in how organizations deploy machine learning in medical and insurance settings.
Your membership also unlocks:
