This month, the MDCG, together with the Joint Artificial Intelligence Board, published a guidance document on how to handle AI-enabled medical devices under the forthcoming AI Act (AIA) in parallel with existing requirements under the MDR/IVDR.
The document introduces the term Medical Device AI systems (MDAIs) and provides a structured overview of how the two regulatory frameworks interrelate. According to the guidance, MDAIs are classified as high-risk AI systems if they serve a medical purpose or are considered a safety component. This means that such products require dual compliance: they must meet all MDR/IVDR obligations and comply with the AI Act’s requirements on aspects such as data governance, transparency, human oversight, cybersecurity, and lifecycle risk management.
Key takeaways include:
- Single technical documentation: One comprehensive documentation set is expected, integrating both MDR/IVDR and AIA requirements.
- Post-market monitoring obligations: Manufacturers must monitor not only clinical performance but also AI-specific behaviours such as unintended interactions or bias.
- Conformity assessment: AIA requirements must be embedded in the MDR/IVDR assessment process.
- Change management: Substantial modifications to the AI logic may trigger reassessment – unless pre-determined change plans are in place.
- Training obligations: Users of MDAIs (e.g. clinicians) must receive adequate training to understand and safely operate the system.
For MedTech manufacturers, this means that the regulatory bar for AI-based products will rise further. It is now essential to integrate AI-specific risk considerations early in product development – and to align quality and documentation processes accordingly. Companies planning to develop or bring to market AI-enabled devices should begin preparing for dual compliance now.
See full document here: AIB 2025-1 MDCG 2025-6
At EUMEDIQ, we are here to support you – whether you need guidance on technical documentation, conformity assessment strategies, or the integration of AI-related requirements into your quality management system. Drop us a mail via info@eumediq.eu.