How the FDA regulates AI

The FDA treats imaging AI tools as medical devices and is building a new framework for AI that learns and evolves.

In the United States, AI software used to make medical decisions is regulated as a medical device. Manufacturers submit evidence of safety and effectiveness, and the FDA authorises the product for a specific intended use before marketing.

Traditional devices are frozen once approved; AI can keep learning. The FDA is therefore developing a “predetermined change control” framework — manufacturers specify in advance what updates an AI may make and validate them through a structured quality process without a full new review each time.

Key FDA concepts

Intended use defines what the AI may do
Pre-market review validates safety and effectiveness
Predetermined change control for evolving AI
Post-market surveillance for real-world safety

In brief

The FDA regulates imaging AI as medical devices.

Approval covers a specific intended use.

A new framework manages AI that keeps learning.

Keep learning

FDA-cleared systemsAI Hub Clinical validation and evidenceAI Hub CE-marked systems in EuropeAI Hub