Researchers flag risks in EU AI changes

A research paper by Hannah van Kolfschooten, Barry Solaiman and Daria Onitiu examines how recent European Union policy proposals could affect safeguards for medical AI under the EU AI Act. The study focuses on changes linked to broader simplification initiatives.

According to the authors, the reforms could maintain the classification of AI-enabled medical devices as high risk while removing key obligations tied to that classification. These include requirements on data governance, risk management and human oversight.

The paper argues that this shift would separate risk classification from the safeguards that give it practical meaning. It suggests that reliance may move back towards existing medical device laws without equivalent AI-specific protections.

The authors warn that such changes could weaken oversight, increase legal uncertainty and affect patient safety where AI systems influence clinical decisions in the European Union.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!