If you have a Medicare Advantage plan, you have almost certainly run into prior authorization at some point. It is the process where your insurance company must approve a procedure, specialist visit, medication, or piece of durable medical equipment before you can actually receive it. For millions of Medicare Advantage enrollees, prior authorization has been a source of genuine frustration — delays that stretch into weeks, denials that seem disconnected from what your doctor recommends, and a paperwork burden that falls on both patients and physicians. Now, the Centers for Medicare and Medicaid Services is testing whether artificial intelligence can make that process faster and more consistent. But a growing number of lawmakers are not willing to take the insurance industry's word for it that the technology is working in patients' favor, and the outcome of that debate will directly shape the care you receive from your Medicare Advantage plan in the years ahead.
The core concern is straightforward. AI systems are trained to approve or deny claims based on patterns in historical data, and if those patterns reflect a systematic bias toward denial — or if the algorithm is optimized to reduce insurer costs rather than improve patient outcomes — then automating prior authorization could make a bad situation worse at enormous scale. Democratic members of Congress have formally pushed CMS to release granular outcome data from the pilot program, including approval rates, denial rates, appeal overturn rates, and a direct comparison of AI-assisted decisions against those made by human reviewers. Without that transparency, there is no way for beneficiaries, physicians, or independent researchers to know whether the technology is actually serving patients.
To understand why this matters, it helps to know how widespread prior authorization has become inside Medicare Advantage. Unlike Original Medicare — Parts A and B — which generally pays for any medically necessary service a physician orders, Medicare Advantage plans are private insurance alternatives that are permitted to require prior authorization for a broad range of services. That includes inpatient hospital stays, skilled nursing facility care, home health services, durable medical equipment such as wheelchairs and CPAP machines, and many specialty medications. The stated intent was to prevent unnecessary or duplicative care. In practice, prior authorization has frequently delayed or blocked care that physicians deemed medically necessary.
According to CMS.gov data, Medicare Advantage plans processed approximately 46 million prior authorization requests in 2022 alone. Of those, roughly 2 million were denied at the initial review stage. Many of those denials were eventually overturned on appeal, but the appeals process itself takes time — time that a 72-year-old waiting for a hip replacement or a cancer patient waiting for a chemotherapy approval may not have. A 2022 report from the HHS Office of Inspector General found that Medicare Advantage plans sometimes denied prior authorization requests that met Medicare's own coverage rules, meaning the denials were not clinically justified under the standards that would have applied in Original Medicare. That finding set the stage for the current scrutiny of AI-assisted review systems.
The AI pilot works like this: when a physician submits a prior authorization request, an algorithm reviews the clinical documentation against a set of criteria — diagnosis codes, treatment guidelines, the patient's claims history — and either approves the request automatically, flags it for human review, or generates a denial recommendation. Proponents argue this speeds up approvals for routine, clearly appropriate requests, freeing human reviewers to focus on complex cases. Critics argue that the criteria embedded in these algorithms may be too rigid, may not account for individual patient circumstances, and may systematically disadvantage patients with complex or atypical conditions who do not fit neatly into the data patterns the system was trained on.
The legislative push for more data is not political theater. Federal law already requires Medicare Advantage plans to make prior authorization decisions within 72 hours for urgent requests and 14 calendar days for standard requests. But those timelines govern the final decision — the internal mechanics of how that decision is reached, and what role an AI system plays in it, have been largely opaque to the public. Lawmakers are asking CMS to require plans using AI-assisted review to disclose the algorithm's role in each decision, track whether AI-assisted denials are overturned at higher or lower rates than human-generated denials, and publish that data so researchers and patient advocates can analyze it independently.
For beneficiaries currently enrolled in Medicare Advantage, the practical reality is this: your appeal rights do not change based on whether a human or an algorithm reviewed your request. If your plan denies a prior authorization, you are entitled to a written explanation that includes the specific clinical reason for the denial and clear instructions on how to appeal. You can request an expedited appeal if your health condition is urgent, and your plan must respond within 72 hours. If the plan upholds the denial internally, you can escalate to an independent review organization — a third party with no financial relationship to your insurer — and studies consistently show that beneficiaries who pursue appeals have a meaningful chance of having denials reversed. The HHS Office of Inspector General has found that in some analyses, more than 75 percent of appeals that reach the independent review stage are decided in the beneficiary's favor, which makes it well worth pursuing rather than accepting a denial.
Before filing a formal appeal, ask your doctor's office about the peer-to-peer review option. This is a direct phone call between your physician and the plan's medical reviewer, and it frequently resolves denials that would otherwise require a multi-week appeals process. It costs you nothing, takes no action on your part, and can be faster than any formal appeal pathway. Many physicians' offices are familiar with this process but may not mention it unless you ask specifically.
CMS finalized a rule in 2024 that placed new restrictions on Medicare Advantage prior authorization practices. Under that rule, plans must establish a list of items and services that do not require prior authorization, must ensure their prior authorization criteria are based on current evidence-based medical guidelines, and must review those policies annually. The rule also requires continuity of care: if you switch Medicare Advantage plans, your new plan generally cannot require you to restart the prior authorization process for an ongoing course of treatment during a transition period. These protections apply regardless of whether AI is involved in the review process.
Data Snapshot: According to CMS.gov data, there were 3,959 Medicare Advantage plans available to beneficiaries nationwide in 2024, reflecting the enormous scale and diversity of the Medicare Advantage market. Plan policies on prior authorization vary significantly across those offerings — some plans require it for dozens of service categories, others for relatively few. CMS also reported that Medicare Advantage enrollment reached approximately 33.8 million beneficiaries in 2024, meaning more than half of all Medicare enrollees are now in plans where prior authorization policies directly affect their access to care. The variation in how aggressively different plans use prior authorization is one of the most important — and least discussed — factors when choosing a plan during open enrollment.
If you are shopping during the Annual Enrollment Period, which runs October 15 through December 7 each year, or the Open Enrollment Period, which runs January 1 through March 31, it is worth specifically investigating a plan's prior authorization requirements before you enroll. Medicare.gov's Plan Finder tool at medicare.gov/plan-compare allows you to compare plans side by side, and you can call 1-800-MEDICARE (1-800-633-4227) to ask specific questions about a plan's utilization management policies. Ask the representative how many service categories require prior authorization and whether the plan uses automated review tools — plans are required to answer these questions.
The broader policy tension here runs through all of Medicare Advantage. Plans are paid a fixed amount per enrollee by the federal government, which means every dollar spent on care reduces insurer margin. That financial structure creates an incentive — not necessarily acted upon in bad faith, but structurally present — to find reasons to delay or deny care. An AI system trained on historical approval and denial data from a plan that has been systematically over-denying care could perpetuate and accelerate those patterns at a speed and scale no human review team could match. That is precisely why the demand for outcome data is so important: you cannot evaluate whether AI is helping or hurting without knowing what it is actually doing to real patients' claims.
If prior authorization delays are affecting your care right now, there are several concrete steps to take. First, ask your doctor's office to submit the prior authorization request with as much supporting clinical documentation as possible — the more evidence of medical necessity in the initial submission, the lower the chance of a denial. Second, pursue the peer-to-peer review option described above. Third, if you need to file a formal appeal, contact your State Health Insurance Assistance Program, known as SHIP, which provides free, federally funded counseling in every state. SHIP counselors can walk you through the appeals process step by step, help you understand the specific language to use in your appeal, and identify whether your denial may have violated CMS coverage rules. You can find your local SHIP counselor at shiphelp.org or by calling 1-800-MEDICARE.
Algorithms do not have medical licenses. They cannot weigh the human cost of a delayed surgery the way a physician can, and they cannot account for the particular circumstances of your case that fall outside the data patterns they were trained on. The lawmakers pushing for public outcome data are asking a reasonable question: if an AI system is making decisions that affect whether a 68-year-old gets her knee replacement or a 75-year-old gets his cardiac stent, the public deserves to know how well that system is performing — and in whose interest. Until that data is public, beneficiaries should know their rights, use every tool available to challenge denials, and pay close attention to prior authorization policies when choosing or renewing a Medicare Advantage plan each fall.