AI could help health plans simplify prior auth, comply with regs

By Mike Selvage, senior manager, and Albert Yu, managing director, Deloitte Consulting LLP

Federally regulated health plans—including Medicare Advantage, Medicaid, and coverage sold through HealthCare.gov—are now required to make standard prior-authorization decisions within seven calendar days, down from the current 14 days. Health plans that fail to comply with the rule, which went into effect on January 1, could face substantial penalties, according to the Centers for Medicare & Medicaid Services (CMS)ⁱ. The new rule does not change the current 72-hour timeframe for expedited prior-authorization decisions.

Prior authorization can be an expensive and time-consuming process for health plans. It often requires their clinical staff to sift through patient records and clinical information—sometimes hundreds of pages—to determine whether a request is medically appropriate. In some cases, a health plan’s clinical staff might need to contact the physician for more information. In addition, if a physician isn’t familiar with a health plan’s requirements for medical necessity, incomplete information might be sent to the health plan, adding to the review timeⁱⁱ.

Delayed prior-authorization decisions can also have a negative impact on patient care and create operational challenges for physicians and health systems that might need to reschedule procedures or tests. Beginning March 31, 2026, CMS will require health plans to publicly report their average turnaround times for prior-authorizations—along with denial, appeal, and overturn rates—via their websitesⁱⁱⁱ. Some consumers could use this information to compare or evaluate health plans. The Department of Health and Human Services (HHS) and CMS finalized the requirements in an Interoperability and Prior Authorization Final Rule that seeks to enhance the electronic exchange of data^iv.

Generative and agentic artificial intelligence (AI) have the potential to reduce the administrative workload associated with prior authorizations, enhance the accuracy of determinations, and reduce human error. The technology could be trained to mine medical records, clinical documentation, and the health plan’s own criteria to validate evidence of medical necessity and auto-approve requests. AI could also help health plans comply with the new regulations from CMS. Clinical oversight will still be needed.

Most health plan execs see AI as an effective tool
Most surveyed health plan executives (93%) expect AI to add value to their companies by automating prior authorizations, according to Deloitte’s 2026 US Health Care Outlook Survey. Respondents also selected AI as the trend most likely to impact their organization’s strategies this year. The level of investment in AI varies by health plan, and large, national carriers tend to be more invested than many small and regional carriers, according to Deloitte’s survey results. At this point, however, there are a limited number of scaled off-the-shelf AI solutions being marketed by vendors to conduct prior authorizations across all service categories. Moreover, some health plans might have limited experience in AI or lack the clinical perspective needed to train AI models and validate the decisions^v.

In addition to AI, the CMS regulations could prompt health plans to consider other strategies to improve turnaround times. For example, a health plan might stop requiring prior authorization for low-cost services and those that have low denial rates (such as home health visits), and procedures that have low clinical risk. Some health plans already use a strategy known as gold-carding where claims are approved for providers that have a history of closely adhering to the health plan’s clinical guidelines and have had few claims denied^vi.

Medicare pilot will use AI for prior authorization
On January 1, CMS launched a pilot allowing AI to support prior-authorization decisions for certain medical services in traditional Medicare. The Wasteful and Inappropriate Service Reduction (WISeR) program was rolled out in six states (Arizona, Ohio, Oklahoma, New Jersey, Texas, and Washington). AI will be used for decisions related to procedures including skin and tissue substitutes, electrical nerve-stimulator implants, and knee arthroscopy. CMS explains that those services are vulnerable to “fraud, waste, and abuse.” While other procedures may be added to the list in the future, inpatient and emergency procedures will not be subject to the AI assessment, according to CMS. The pilot program was developed to reduce wasteful, “low-value” services, according to a CMS statement^vii. The model excludes inpatient-only services, emergency services, and treatments that would pose a significant risk to patients if delayed. The agency stressed that clinicians must be involved throughout the process, and denials will only be made by licensed clinicians. Depending on what is learned from the WISeR pilot, AI could become an integral part of the prior-authorization process.

Conclusion
Medical policies used in prior authorizations are becoming increasingly nuanced. At the same time, the medical files (PDFs) that are submitted to health plans have become longer and more detailed due to the ease at which they can be generated by electronic health records systems.

Adding clinical staff to comply with CMS’s updated prior-authorization regulations is probably not a sustainable model. While AI adoption varies across the industry, its ability to mine medical records, validate clinical criteria, and surface relevant insights could position it as an important tool for meeting new turnaround mandates and reporting standards. If used correctly, AI could also be a lever for increased quality and consistency in prior-authorization reviews. Along with helping to reduce administrative costs, AI could also result in faster prior-authorization decisions. But the technology should be viewed only as an augmentation for clinical staff.

Latest news from @DeloitteHealth

Endnotes

ⁱPrior authorization Final Rule, CMS, February 8, 2024
ⁱⁱPrior authorization: How it evolved, why it burdens physicians and patients, and the promise of AI, Medical Economics, April 24, 2025
ⁱⁱⁱPrior authorization Final Rule, CMS, February 8, 2024
^ivAdvancing Interoperability and Improving Prior Authorization Processes, Federal Register, February 8, 2024
^vArtificial intelligence adoption challenges from health care providers’ perspectives, Safety Science/Elsevier, January 2026
^viPrior authorization reform gains momentum in states, MultiState, August 14, 2025
^viiCMS launches new model to target wasteful, inappropriate services, CMS, June 27, 2025

This publication contains general information only and Deloitte is not, by means of this publication, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This publication is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional advisor.

Deloitte shall not be responsible for any loss sustained by any person who relies on this publication.

Return to the Health Forward home page to discover more insights from our leaders.