AI is now embedded in health insurance prior authorization systems, and 61 percent of physicians surveyed by the American Medical Association in 2025 believe it will increase wrongful denials of medically necessary care.
Prior authorization already delays and derails treatment. Doctors flag it as a source of patient abandonment of recommended therapies. The system exists to curb overuse and redirect spending toward cost-effective alternatives, but it has become a bottleneck that insurers and AI vendors now claim automation can fix. The argument is straightforward: AI sorts large volumes of claims faster than humans, theoretically clearing unambiguous approvals without delay. The problem is that faster is not the same as more accurate.
The full article is worth reading for how it maps the gap between the theoretical efficiency gains of AI-driven authorization and the documented pattern of increased denials. The tension between insurer cost control and physician judgment is not new, but AI gives insurers a scalable tool to operationalize that tension at volume.
[READ ORIGINAL →]