AI-Based In-House Medical Devices at a Regulatory Crossroads: Beneficial Regime or Legal Loophole?

Abstract The Medical Device Regulation (MDR) establishes an in-house exemption regime that allows medical devices, manufactured and used exclusively within health institutions in the European Union (EU), to be put into service with reduced regulatory requirements. This exemption excludes the need for a conformity assessment by a notified body, provided that the general safety and performance requirements are met. As a result, Artificial Intelligence (AI)-based in-house medical devices, despite their potential risk classification, do not fulfil the cumulative conditions outlined in Article 6 of the Artificial Intelligence Act (AI Act) necessary for categorisation as high-risk AI systems. This regulatory disconnect reveals a significant gap in the alignment between the MDR and the AI Act, raising concerns about regulatory oversight for AI-based in-house medical devices. Academia has already explored the application of the AI Act on medical devices and the interplay between the MDR and the AI Act. What remains insufficiently examined, however, is the classification under the AI Act of AI-based in-house medical devices. This paper examines this issue and analyses whether the low-risk classification may amount to a legal loophole. Ultimately, it also offers recommendations on how to safeguard innovation while simultaneously reinforcing safety.