Will AI fix prior authorization—or make it worse?

2026-07-18 · Source: AI - Ars Technica · Field: Health & Wellbeing — Healthcare Systems & Policy, Clinical Care & Medical Practice, Medical Devices & Health Technology · Depth: Fundamental Awareness, medium

Summary

The healthcare system faces significant challenges with prior authorization, a process for pre-approving medical care that often leads to care delays and patient treatment abandonment. While intended to curb overuse, a 2025 American Medical Association survey found 61 percent of physicians fear AI will worsen wrongful denials. A 2025 Commonwealth Fund survey revealed one in five privately insured working-age adults experienced coverage denials, with 41 percent facing delayed care and over a quarter seeing their health worsen. The Trump administration is expanding AI use through the WISeR model, a demonstration project running through December 2031 in six states, aiming to reduce waste and fraud in original Medicare. However, critics, including health policy analysts and lawmakers, express concern that WISeR, which allows vendors to profit from "averted expenditures," could increase denials and administrative burden, despite insurers' pledges to streamline processes and reduce prior authorization requests by 11 percent between June 2025 and April 2026.

Key takeaway

For healthcare administrators and policymakers evaluating AI integration into prior authorization, recognize that current implementations like the WISeR model risk increasing patient care denials and administrative burdens. You should prioritize systems that mandate transparent clinical reasoning for denials and ensure robust human oversight of AI algorithms. Focus on solutions designed to expedite appropriate approvals, not merely to accelerate denials, to safeguard patient access and trust.

Key insights

AI's application in prior authorization risks exacerbating care denials and patient burden despite its potential for efficiency.

Principles

• Prior authorization balances cost control with patient access.
• AI in healthcare requires transparency and clinical oversight.
• Financial incentives can distort AI-driven denial decisions.

Method

The WISeR model combines machine learning with human clinical review to evaluate services vulnerable to overuse, fraud, and abuse in original Medicare, aiming to reduce unnecessary procedures.

In practice

• Advocate for detailed clinical reasoning for denials.
• Demand transparency in AI algorithms for coverage.
• Monitor AI systems for unintended denial increases.

Topics

• Prior Authorization
• AI in Healthcare
• WISeR Model
• Medicare Policy
• Insurance Denials
• Healthcare Regulation

Best for: Executive, CTO, VP of Engineering/Data, AI Ethicist, Policy Maker, Domain Expert

Related on AIssential

• THE BRONX NEEDS REAL NURSES, NOT AI! — Replacing experienced nurses with AI for critical utilization review tasks risks patient care quality, job security, and private data integrity.
• How Data Analytics Is Changing Healthcare Risk Management — New Medicare Advantage regulations mandate a shift to compliance-first coding, prioritizing audit-defensible documentation over revenue maximization.
• AI policy must fail gracefully — AI policy must be designed for graceful degradation, distributing power beyond the executive branch to ensure resilience against political shifts.
• Medical records could be revealed by AI training-data vulnerability — Medical AI models are vulnerable to membership inference attacks, revealing sensitive patient data, with risk increasing with model capacity.
• 5 Security Mistakes That Kill Healthcare AI Startups — Healthcare AI startups must integrate security as a core architectural decision, not a compliance afterthought, to avoid catastrophic failures.
• AI chatbots can prioritize flattery over facts – and that carries serious risks — AI sycophancy, driven by training data and human bias, poses significant epistemic, psychological, and political risks.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI - Ars Technica.