EU Clarifies That High-Risk AI Rules Will Hit Many Medical Devices Later Than 2026
Updated European Commission guidance on the AI Act makes clear that many AI systems embedded in regulated products such as medical devices are classified as high-risk, but their specific obligations are set to apply from August 2, 2027. The timing matters for manufacturers trying to map MDR, product liability and AI Act requirements into one compliance strategy.
Europe’s healthcare AI regulatory picture became clearer this year as the European Commission expanded its public guidance on the AI Act. In a February 2026 FAQ update, the Commission reiterated that AI systems embedded in regulated products, including medical devices, are generally considered high-risk when they are subject to third-party conformity assessment under sector-specific legislation.
For healthcare companies, the most important detail is timing. While most AI Act rules start applying on August 2, 2026, the Commission says obligations for high-risk AI systems embedded in regulated products listed in Annex II, including medical devices, apply 36 months after entry into force, on August 2, 2027. The Commission also noted that a November 2025 Digital Omnibus proposal could further adjust the timeline by linking it to the availability of support measures such as standards and guidance.
That matters because medtech makers are not dealing with the AI Act in isolation. The Commission’s healthcare AI overview also ties the new regime to the European Health Data Space and the Product Liability Directive, underscoring that software and AI providers may be treated as manufacturers and face defect-based liability questions if systems continue learning after deployment. In other words, compliance is becoming a multilayered product-governance problem, not just a documentation exercise.
The net effect is mixed. On one hand, the delayed application date for embedded high-risk systems gives medical-device developers more runway. On the other, it confirms that healthcare AI in Europe is moving toward a tougher, more formal oversight model. For companies building clinical decision support, in vitro diagnostics or AI-enabled medical software, 2026 is increasingly looking like a preparation year rather than a finish line.