Why Prior Authorization is Already a Headache (and Why AI Might Not Help)
Prior authorization (PA) is a notorious bottleneck in healthcare, a system meant to control costs but often perceived by providers and patients as an arbitrary barrier to necessary care. The prospect of AI prior authorization entering this already contentious process elicits a mix of cautious optimism and deep skepticism.
Many healthcare providers express deep frustration, viewing PA as a system designed to deny care. There's a hope that AI could ease the paperwork, with some medical assistants and nurses exploring AI tools to automate calls and draft appeals, aiming to reduce time spent on hold. However, the implementation of AI in prior authorization also brings significant concerns about its impact on patient care and administrative burdens.
A common concern is that AI will just automate and worsen the current problems. This could lead to a more sophisticated "AI vs. AI" battle between insurers and providers, where patients suffer from increased denials and delays. Understanding the nuances of AI prior authorization is crucial for all stakeholders.
The Federal Push: AI for "Waste Reduction"
Most health insurers already use automated AI systems for PA requests. Approximately three out of four plans in individual and group markets use AI for PA approvals, and 8–12% use it to support PA denials. While automated AI systems for PA requests are not new to most health insurers, the federal government's direct involvement marks a significant shift in the landscape of AI prior authorization.
In early 2026, the Centers for Medicare & Medicaid Services (CMS) launched the Wasteful and Inappropriate Service Reduction (WISeR) Model. This pilot program, active in Arizona, New Jersey, Ohio, Oklahoma, Texas, and Washington, contracts with third parties to use AI to review authorization requests for select services in traditional Medicare.
CMS says its goal is to target services vulnerable to fraud, waste, and abuse. Initially, it focuses on things like skin and tissue substitutes, electrical nerve stimulator implants, and arthroscopy for knee osteoarthritis. This federal initiative highlights a growing reliance on AI prior authorization for cost control.
Crucially, under WISeR, physicians aren't required to submit prior authorization requests *before* providing these targeted services. Instead, claims are subject to medical review *after* the service is delivered to ensure it met Medicare coverage, coding, and payment criteria prior to payment. This shifts the timing but keeps the AI review in play, potentially impacting how providers approach care delivery and documentation for AI prior authorization systems.
When "Efficiency" Means More Denials
While proponents, including some insurers and government agencies like CMS, point to AI's potential to streamline processes and cut paperwork, physician organizations and lawmakers express serious concern regarding AI prior authorization.
The American Society of Nuclear Cardiology (ASNC), among 35 physician organizations, opposed the WISeR model in 2025. Their worry, and that of groups like the American Medical Association, is that the WISeR payment structure rewards vendors for authorization denials. If a system is designed to find "waste" and its operators are paid based on what they deny, this creates a perverse incentive where operators are rewarded for authorization denials, directly impacting patient access to care through AI prior authorization.
This is akin to a a quality control system that is rewarded for finding defects, even if those "defects" are necessary components. The system becomes efficient at its given task, but the outcome might not be what's best for the patient. This fundamental flaw in incentive structure is a key argument against unchecked AI prior authorization.
Lawmakers are also pushing back. The House of Representatives Appropriations Committee approved legislation that would bar Health and Human Services (HHS) federal funding from being used to implement any model that applies prior authorization in traditional Medicare, including AI-based models like WISeR. Companion bills have also been introduced in both the House and Senate to stop WISeR, reflecting ongoing legislative concern, though the bill is unlikely to gain traction in the current Congress. These legislative efforts underscore the contentious nature of AI prior authorization.
The fear is that WISeR could set a precedent, expanding AI-driven prior authorization to more HHS programs and potentially targeting advanced diagnostic imaging in future years. This potential expansion raises alarms about the future of healthcare access and the role of AI prior authorization.
States Try to Build Guardrails, Feds Push Back
Recognizing the risks, many states are trying to fill the regulatory gap. They're creating AI-specific laws for high-risk uses, often establishing new nondiscrimination protections and clarifying obligations for transparency and consumer safeguards in the context of AI prior authorization.
For example, Colorado's Consumer Protections in Interactions with Artificial Intelligence Systems Act, set to be implemented in June 2026, applies to AI used in healthcare decisions, including utilization decisions. It guarantees bias protections, requires plans to disclose data and methodologies, and gives individuals the right to appeal an AI-generated healthcare decision. This represents a proactive step towards responsible AI prior authorization.
Other states, like Texas (with 2025 legislation), Arizona, and Maryland, have adopted similar laws prohibiting the use of AI as the *sole* basis for a medical necessity denial, mandating human oversight. These state-level initiatives aim to ensure that AI prior authorization remains a tool to assist, not replace, human judgment.
However, a federal Executive Order on AI, titled "Ensuring a National Policy Framework for Artificial Intelligence," aims to limit states' ability to enact and enforce their own AI safeguards. It directs the Department of Justice to identify and challenge state laws conflicting with federal AI policy or deemed "onerous or excessive." This federal stance creates tension with state efforts to regulate AI prior authorization.
This creates real pressure on states, potentially pushing them toward less meaningful reforms and harming patients who lack protection against opaque algorithmic systems and difficult-to-challenge denials in AI prior authorization. The balance between federal and state regulation will be critical for the future of AI in healthcare.
What This Means for Patients and Providers
The implications of AI in prior authorization are unfolding through legislative efforts to halt or regulate these models, and in states actively developing regulations to ensure transparency, fairness, and the continued involvement of human clinicians in AI-assisted coverage decisions. Both patients and providers need to be aware of these developments.
While AI offers potential, the fundamental challenge lies not with the technology itself, but with how it's designed and deployed within a healthcare system already grappling with significant problems. An AI model works by finding patterns in vast datasets. If that data reflects existing biases or incentives to deny care, the AI will learn and amplify those patterns. AI is a powerful tool, yet its effectiveness and fairness are inherently tied to the quality of its training data and the specific parameters it's programmed to follow, especially in sensitive areas like AI prior authorization.
My Take on AI in Prior Authorization
AI has the potential to streamline some administrative tasks in prior authorization, like quickly identifying complete paperwork or flagging obvious errors. That could genuinely save time for busy medical staff, improving efficiency without compromising care.
But if AI is used to make final denial decisions without solid human oversight, transparency, and clear appeal processes, it will likely worsen patient access to care. The current federal push to expand AI in PA, coupled with efforts to preempt state-level protections, is a serious concern that could undermine the benefits of AI prior authorization.
We need AI that *assists* clinicians and patients, not AI that acts as an unchecked arbiter of care. Patients and providers must closely monitor how these systems are implemented and advocate for strong human review and appeal rights. The pursuit of fair prior authorization continues, now with the added complexity of artificial intelligence shaping its future.
