EU discussions on MDR/IVDR: TÜV Association warns against sector exit

Calling for AI testing while scrapping AI rules? TÜV Association warns against a sector exit for medical devices from the AI Act through the back door of the MDR reform.

Humanoider Serviceroboter in einem hellen Patientenzimmer eines Krankenhauses. Der weiß-schwarze Roboter trägt ein Tablett mit Medikamenten und medizinischem Zubehör. Im Hintergrund stehen ein Krankenbett, ein Rollstuhl und medizinische Geräte.
©Getty/Unsplash

Berlin, 30. September 2026 – Ahead of the meeting of EU health ministers in Dublin, TÜV Association warns against using the reform of the Medical Devices Regulation (MDR) and In Vitro Diagnostic Medical Devices Regulation (IVDR) to undermine the AI Act’s safety requirements for AI-enabled medical devices. Only in May, the Council and the European Parliament agreed in the context of the AI Omnibus to keep medical devices fully within the scope of the AI Act. The MDR reform now puts this decision back on the table. “The MDR reform must not be used as a pretext to scrap AI rules through the back door that have nothing to do with the reform itself,” says Mathea Essinger, Head of TÜV Association’s Brussels Office. “When AI is used to make diagnoses or recommend treatments, AI safety requirements are not red tape – they are about protecting patients. A sector exit would mean that the Commission would first remove these requirements, only to reintroduce them, potentially in a less stringent way, even though they are already laid down in the AI Act. That is the opposite of cutting red tape.”

Contradictory signals on AI safety

The debate comes at a time when AI risks are receiving intense international attention. Just a few days ago, German Chancellor Friedrich Merz, European Commission President Ursula von der Leyen and other heads of state and government called in a joint declaration for mandatory testing and independent evaluations of powerful AI models. “Europe cannot say one thing on AI safety and do another,” Essinger says.

High-risk requirements for AI in medical devices would no longer apply directly

The European Commission’s proposal would move medical devices in Annex I of the AI Act from Section A to Section B. Such a sector exit could mean that key high-risk requirements would no longer apply directly to AI-enabled medical devices - or only in a weakened form. These include requirements on risk management, data quality, transparency, human oversight and robustness.

It remains unclear to what extent the Commission would reintroduce these requirements, with a risk that they could be weakened in the process, potentially reducing the level of patient protection. TÜV Association calls on EU Member States to reject the proposed move to Section B, keep AI-enabled medical devices fully within the scope of the AI Act and ensure that the existing requirements are implemented in a practical and workable manner.