The EU AI Act and Healthcare: What Clinicians Need to Know

The European Union’s Artificial Intelligence Act — the world’s first comprehensive legal framework for AI — entered into force on August 1, 2024. For healthcare, the implications are substantial. Medical AI systems used in clinical decision-making are explicitly classified as high-risk AI under Annex III of the Act, triggering a conformity assessment process that parallels — and in some respects exceeds — the CE marking requirements under the Medical Device Regulation (MDR).

The High-Risk Classification: What It Covers

Under the EU AI Act, high-risk AI systems in healthcare include AI intended to: assist clinical decision-making for diagnosis or treatment; support triage or prioritization of patients; monitor physiological parameters in real time; and predict disease progression. This definition covers the vast majority of commercially deployed clinical AI tools — including diagnostic imaging AI, sepsis prediction algorithms, risk scoring systems, and AI-enabled patient monitoring platforms.

High-risk classification triggers mandatory requirements for: risk management systems, data governance documentation, technical documentation (substantially similar to MDR technical files), transparency and logging requirements, human oversight mechanisms, and accuracy and robustness specifications.

The Timeline

The EU AI Act’s implementation follows a phased schedule:

February 2025: Prohibited AI practices rules apply. This bans social scoring, real-time biometric surveillance (with narrow exceptions), and AI that exploits psychological vulnerabilities — applications that are largely not relevant to clinical healthcare but affect administrative AI in health systems.

August 2025: GPAI (general-purpose AI model) rules and governance provisions apply. This affects large language models used in clinical applications — including AI-assisted documentation, clinical summarization, and diagnostic support chatbots.

August 2026: High-risk AI provisions fully apply. This is the critical deadline for clinical AI tools — when conformity assessments, technical documentation, and registration requirements become mandatory.

The Intersection with MDR

Many AI medical devices in Europe are already regulated under the MDR (EU 2017/745) and In Vitro Diagnostic Regulation (IVDR). The EU AI Act introduces a layered compliance requirement: AI systems that are already regulated as medical devices must comply with both MDR/IVDR requirements and relevant EU AI Act obligations.

The European Commission has acknowledged the potential for compliance duplication and is developing guidance on how the two regulatory frameworks interact — but as of early 2026, the detailed harmonization guidance remains incomplete. This creates genuine compliance uncertainty for manufacturers of regulated medical AI.

What Notified Bodies Are Doing

EU notified bodies — the accredited third-party organizations that perform conformity assessments for medical devices — are expanding their AI assessment capabilities. BSI, TÜV SÜD, and other major notified bodies have announced AI-specific assessment programs and are recruiting AI expertise to support conformity review of high-risk medical AI systems.

The bottleneck is capacity. The number of AI medical devices requiring conformity assessment under the EU AI Act dwarfs current notified body capacity. Industry associations have raised concerns that the August 2026 deadline may produce a significant compliance backlog.

Practical Implications for Clinicians and Health IT Teams

For European healthcare organizations deploying AI tools, the EU AI Act creates new due diligence requirements. Procurement of high-risk AI systems will require verification that vendors have completed (or are on track to complete) conformity assessments before the August 2026 deadline. Post-market monitoring obligations will require health systems to log AI system performance and report serious incidents to national competent authorities.

The EU AI Act’s transparency requirements also mandate that patients be informed when AI systems are used in decisions that significantly affect their care — a requirement that will necessitate updates to patient consent forms and information materials across European healthcare providers.

Sources: Official Journal of the European Union, EU AI Act, August 2024. European Commission AI Act implementation guidance. EU Medical Device Regulation (2017/745). BSI AI conformity assessment program announcement, 2025.