Can AI untangle prior authorization, or will it deny more patients?

The struggle to obtain insurance pre-approval for recommended medical treatments remains a source of frustration for patients and physicians alike. Delays in the prior authorization process—where insurers verify that a proposed treatment is medically necessary before covering costs—can lead patients to forgo care while waiting for a decision.

The tension between cost control and access#

Prior authorization exists as a brake on unnecessary spending, flagging treatments where cheaper alternatives might suffice. But the American Medical Association and other groups have documented that the system often stalls legitimate care, forcing patients into extended waits or appeals processes that demand even more time.

AI's promise and peril#

Artificial intelligence could theoretically accelerate this bottleneck by rapidly reviewing clinical documentation and approving straightforward cases. Yet physicians are skeptical. A 2025 AMA survey found that 61 percent of doctors worry AI will make it easier to wrongfully deny coverage—shifting the system from a gateway that sometimes delays care to one that more readily blocks it.

The core concern is that AI algorithms, trained on historical claims data, may learn to deny claims at higher rates. "AI should be used to make appropriate care easier to approve, not necessary care easier to deny," health policy analyst Camm Epstein told Undark.

Calls for oversight#

The AMA is pushing for insurers to explain their denial decisions in clinical detail and to provide greater transparency about the algorithms themselves, ensuring doctors retain meaningful oversight.

Meanwhile, the Trump administration has launched a pilot program in six states using AI to trim unnecessary medical costs—a real-world test of whether automation can fix rather than further break an already strained authorization system.