Traditional Medicare has always had one clear advantage over Medicare Advantage in the minds of many beneficiaries who choose it deliberately: a doctor could order a test or procedure and it would simply happen, without the prior-authorization gatekeeping that Medicare Advantage insurers have long used to review and sometimes deny requests before care is delivered. That distinction is now eroding from the inside. Federal regulators launched a demonstration program this year testing AI-powered review of certain procedure requests within original Medicare itself, the exact program structure beneficiaries were choosing traditional Medicare specifically to avoid.

The program, aimed at reducing wasteful or inappropriate procedures, uses artificial intelligence to process authorization requests, and it's already generating the kind of complaint usually associated with Medicare Advantage denials: patients waiting weeks for approval on services that, before this year, never required approval at all. One senator overseeing the program's early results described the AI system as functioning less like a quality check and more like what he called a "denial device."

A Number That Undercuts the Whole Premise

The single most damning data point to emerge from this program's early rollout is an appeals statistic: a federal watchdog report found that 95 percent of denials that were actually appealed ended up overturned. That's not a modest error rate suggesting occasional mistakes in an otherwise sound system. That's a system whose initial output is wrong the overwhelming majority of the time it gets checked, which raises the obvious question of what the AI review is actually filtering for, if not simply generating a default denial that only gets corrected when a patient or provider has the wherewithal to fight it.

The Mechanics of a Delay That Functions Like a Denial Even When It Isn't One

Here's the structural reason a 95 percent overturn rate doesn't actually protect most patients, even though it sounds like the appeals process is working as intended. The appeals process itself typically takes 30 to 60 days to resolve, a window during which the patient simply doesn't receive the service and the provider doesn't get paid for it. For many medical situations, a two-month delay isn't a neutral pause while paperwork gets sorted, it's a real deterioration in whatever condition the procedure was meant to address. Separate litigation against major Medicare Advantage insurers has alleged that some patients died before a final decision on their appeal of an AI-generated denial was ever reached, an extreme but illustrative version of the underlying dynamic: an appeal that eventually succeeds on paper doesn't help a patient whose actual medical need couldn't wait that long.

Why Scale Makes This Worse, Not Better

Medicare Advantage plans alone processed close to 53 million prior authorization determinations in a recent year, and the denial rate on those requests has been climbing as insurers increasingly deploy AI systems that can process authorization requests in seconds rather than the hours or days a human reviewer would need. That speed is precisely what makes an AI reviewer so appealing to a payer trying to manage costs: it can issue an initial denial on a request almost instantly, at essentially no marginal cost, counting on the fact that only a fraction of denied patients will have the time, health literacy, or advocacy support to actually file and pursue an appeal. A 95 percent overturn rate on the appeals that do get filed says nothing about the patients who never appeal at all, and structurally, a fast, cheap, automated denial system doesn't need most people to appeal successfully in order to save the payer money. It just needs most people not to appeal.

What This Means for the Traditional Medicare Advantage

None of this means every prior authorization request is wrongly denied, and some genuinely do prevent unnecessary or duplicative procedures, which is the program's stated goal. But an AI review system whose denials get overturned 95 percent of the time on appeal isn't functioning as a quality filter, it's functioning as a default denial that quietly relies on most people never challenging it, and the traditional Medicare beneficiaries this program now applies to no longer have the protection from that dynamic that used to be one of the program's core selling points. For anyone facing a prior authorization delay or denial under this program, the practical step is treating an initial denial as the starting point of the process rather than the final answer, and appealing immediately rather than assuming the decision reflects an accurate medical judgment.

— John Stone