The FDA AI action plan and what it means for procurement

The FDA's updated framework represents a significant shift in how clinical AI is regulated. For health systems evaluating AI tools, the changes have direct procurement implications.

What changed

The framework introduces a distinction between AI that functions as a medical device (SaMD) and AI that supports clinical decision-making without directly influencing it. This distinction affects which tools require FDA clearance and which are subject to lighter oversight.

More importantly, the framework introduces requirements for continuous monitoring of deployed AI systems. Vendors are now expected to track model performance over time and report significant performance changes to the FDA.

What it means for procurement

First, ask every clinical AI vendor for their FDA clearance status and classification. Some tools that previously operated in regulatory gray areas are now clearly subject to oversight, and if they have not sought appropriate clearance, that is a material risk.

Second, ask vendors how they fulfill their post-market surveillance obligations. A vendor who cannot clearly describe their model monitoring program is either not doing it or does not know they are required to.

Third, build FDA clearance status into your vendor evaluation criteria. It is not sufficient on its own, but a vendor who has sought and received appropriate clearance has at minimum engaged seriously with the regulatory process.