Across the pond in Europe, those in the health care industry are wondering what the European Union’s new Artificial Intelligence Act means for them. The AI Act, provisions of which started going into effect in August 2024 and will be fully applicable in August 2026, sets new rules for AI products, including health-related AI technology. But there’s still some confusion over how the rules will impact the field, where it coexists with rules such as the EU’s Medical Devices Regulation, our POLITICO colleagues in Europe report. How does it work? The AI Act uses a risk-based approach, categorizing products and regulating them based on the risk level: unacceptable, high, limited and minimal. In health care, high-risk devices might include AI-enabled clinical decision support systems, automated insulin delivery and emergency call classification, Anett Molnar, policy adviser at the Healthcare Information and Management Systems Society, or HIMSS, said during a webinar on the topic Tuesday. Limited-risk devices would include chatbots for general health advice, sleep-tracking devices or sensors for assisted living. Products such as AI-enabled hospital administration systems would fit into the minimal-risk category. What’s next: Molnar said to expect guidance from the medical devices coordination group on how to interpret the AI Act rules for devices. But the bottom line is that “there is no such a thing as a dual certification,” Molnar said. Independent entities designated by the EU to assess medical devices, must check that AI devices also adhere to the AI Act in the future, Molnar said. “When there is an application for a new device to enter the market, it will be one notified body looking for compliance both for MDR and both for AI,” she said.
| 
