FDA blocks radiology AI exemption

The FDA rejected Harrison.ai's request to skip premarket clearance for four types of radiology software, including cancer detection and diagnostic tools.

What Harrison.ai wanted

The company proposed exempting products from the 510(k) notification process if a manufacturer already had clearance for a related device and maintained post-market monitoring.

FDA said no to that trade

The agency determined that premarket notification is necessary to ensure safety and effectiveness, rejecting the argument that a manufacturer's track record could replace upfront review.

“the petition and public comments did not demonstrate that premarket notification was unnecessary to provide reasonable assurance of safety and effectiveness”

Version control becomes a money problem

AI models improve rapidly, but clinically meaningful changes require new FDA submissions before hospitals can use them, making each update a regulatory and financial hurdle.

Investors must track clearance status

Lenders and investors financing healthcare AI need to know which product versions have FDA approval, which are under review, and how much revenue depends on products not yet cleared to sell.

Hospitals need version controls too

Health systems must connect the software running in clinical settings to the exact FDA-cleared product version, since newer versions are not automatically authorized for the same uses.