ARTICLE SUMMARY:
BSI artificial intelligence lead Sarah Mathew offered up some answers at a recent gathering about how the EU Artificial Intelligence Act will work in practice. She also provided advice for the medtech sector on preparing for implementation of the law, which will institute new assessment requirements that will kick in for most AI medical devices by 2027.
There are still a lot of information gaps about how the recently enacted EU Artificial Intelligence Act will be carried out in practice, but answers are starting to trickle in as notified bodies, among others, dig into the details of the new law.
The AI Act, enacted in August, will institute new assessment requirements for “high risk” AI systems, including many AI medical devices (AIMDs). It will add challenges in Europe for AIMD makers on top of the complexities that persist with the Medical Device and IVD Regulation, but the medtech industry hopes that it won’t lead to duplicative efforts. The cross-sector law is designed to work in coordination with existing EU regulations. So, in the case of devices and IVDs, notified bodies that already audit and assess companies under the MDR and IVDR will be tasked with incorporating AI Act oversight into one relatively seamless process.
That’s the promise at least. Stakeholders have a bit less than three years to make that promise a reality (AI Act assessment will start to be enforced in 2026 for some systems, but generally it won’t take effect for AIMDs until 2027). Notified bodies will be central to these efforts, and the larger ones, at least, like BSI and TÜV SÜD, say they have already put a lot of time and effort into meeting the challenge.
