Last autumn, I sat across from a mid-sized European medtech manufacturer who told me something that has stayed with me. They had developed an AI-assisted mammography screening tool with strong clinical evidence behind it, one that could detect early signs of breast cancer more accurately and at a greater scale than current screening programmes allow. They were not waiting for the science to catch up. They were waiting for someone to tell them which rulebook applied.

The EU is moving fast on the Digital Omnibus on AI proposal, an initiative intended to streamline and future-proof Europe’s digital rulebook, with key discussions in the European Parliament expected in the coming weeks and a political compromise to follow in the months ahead. It is a welcome step. But the question is whether it goes far enough. In healthcare, a framework that falls short on ambition is not simplification. It is a setback. And the current trajectory, if left uncorrected, points in that direction.

Speed is not the same as readiness

The AI Act’s high-risk obligations should not apply until the conditions for compliance actually exist. Harmonised standards finalised. Notified bodies with real capacity. Regulatory guidance that manufacturers can act on. A two-year implementation window starting only once those conditions are met is not a concession to industry. It is basic respect for how complex regulated systems work.

The reality today is stark. Notified bodies are unevenly prepared across Member States. Harmonised standards remain incomplete. Guidance is patchy and, in some areas, absent entirely. Launching binding obligations into that environment does not accelerate safe AI adoption. It produces paperwork while innovation waits, and patients wait with it.

One designation, not two

The bottleneck is not theoretical. Manufacturers developing AI-enabled medical technologies must navigate both the AI Act and the sectoral framework, Medical Devices Regulation (MDR) or In Vitro Diagnostic Regulation (IVDR). Each pathway has its own requirements, its own timeline, and its own points of failure. The cumulative effect is delay: for manufacturers, for notified bodies, and ultimately for patients who need these technologies now, not in three years.

The current Digital Omnibus proposal enables a single application and unified assessment for dual designation. That is good progress. But a unified process built on top of separate legal requirements is not genuine harmonisation. It is administrative tidying. What is needed is a designation framework that actually reduces the burden, not one that consolidates the forms while leaving the underlying complexity intact. Policymakers know the difference. This is the moment to act on it.

A more coherent approach is already emerging in the ongoing revision of the MDR and IVDR. Instead of forcing companies through two parallel sets of rules, the Commission now proposes to align the AI Act with the medical‑device framework so that AI‑specific requirements are handled through the MDR/IVDR process itself. This reduces duplication, keeps one notified body in charge of the full assessment, and ensures that medical‑AI oversight sits within the system designed for patient safety and clinical effectiveness.

The rulebook problem

The AI Act introduces the concepts of “substantial modification” and “safety component” to determine when high-risk classification and notified body involvement are triggered. Both concepts are defined in the AI Act but have no equivalent in the MDR or IVDR, creating uncertainty for manufacturers about when an AI‑enabled device needs reassessment.

For a policy drafter, this may look like a technical detail to be resolved in guidance. For a manufacturer deciding whether their adaptive AI system requires full recertification, it is a legal grey zone that delays decisions, drives up costs, and in some cases stops products from moving forward entirely. The cumulative costs across hundreds of AI-enabled medical technologies, in time, resources, and delayed patient access, are real and growing.

Aligning definitions across the AI Act and sectoral legislation requires political will, not technical genius. Leaving it to guidance documents to sort out years down the line is a choice. And it is the wrong one. Without regulatory coherence between the AI Act and MDR/IVDR, companies are forced into parallel assessment tracks for a single product, an inefficiency that helps no one, least of all patients.

What is actually at stake

The US and China are not pausing to resolve designation pathway questions. They are building clinical evidence, establishing market presence, and quietly setting the standards the rest of the world will follow.

Europe does not need to match their pace. It needs to be smarter. The Digital Omnibus is a genuine opportunity to make the EU’s regulatory framework for AI in healthcare a source of competitive confidence rather than strategic uncertainty. What that requires is straightforward, even if achieving it is not: legal certainty, definitions that align across legislative frameworks, designation pathways that function as intended, and timelines grounded in operational reality rather than political convenience.

None of this is radical. It is the minimum needed to ensure that Europe’s regulatory ambition translates into outcomes that actually reach patients. Right now, that gap is wider than it should be, and the window to close it is narrowing.

Alexander Olbrechts, Director Digital Health & Medtech Value, MedTech Europe