Why it matters
Under the EU AI Act, AI in regulated medical devices is generally treated as high-risk, adding obligations to those in the Medical Device Regulation. Providers that deploy these tools also have responsibilities for their use and oversight.
The AI Act and the Medical Device Regulation apply together. Manufacturers carry most obligations, and hospitals and clinics that deploy AI must make sure it is used as intended, with trained staff and monitoring.
By the numbers
~1,450AI-enabled medical devices on the FDA list by the end of 2025 1
76%of them are in radiology 1
2023
2024
2025
Source: AuntMinnie, Radiology dominates thirty years of FDA AI device approvals (analysis of the FDA AI-enabled device list), 2026
What's changing
- Manufacturers must meet requirements for data governance, transparency and human oversight.
- Deployers must use systems according to their instructions and monitor how they operate.
- Post-market surveillance and incident reporting continue to apply.
How leaders are responding
- Keeping an inventory of AI tools in clinical use.
- Defining clinical oversight and escalation procedures.
- Including AI requirements in procurement.
What good looks like
- An inventory of AI tools classified against medical device and AI Act rules
- Post-market monitoring of performance in your own patient population
- Procurement terms that require vendor transparency and updates
What to watch
- Harmonised standards and guidance for AI in medical devices.
- The AI Act's phased application dates.
- Procurement frameworks for AI in health systems.
Where to start
Assign a clinical owner to each AI tool in use, responsible for oversight, incident reporting and staff training.
Questions for your leadership team
- Which AI tools do we use, and how are they classified?
- Who oversees their performance in practice?
- What evidence do we ask vendors to provide?
