CSOAI - healthcare + EU AI Act
AI in healthcare and the EU AI Act
Clinical AI is squarely in the EU AI Act's high-risk tier. Transparency duties land 2 Aug 2026; the full high-risk regime phases in by Dec 2027. Here is what applies to medical AI - and how to get ready now.
High-risk uses in your sector
- !AI for medical diagnosis, triage, or screening
- !AI components inside medical devices (MDR / IVDR overlap)
- !Patient risk-stratification and resource allocation
- !AI-driven clinical decision support
What you must do
- +Human oversight of every clinical decision
- +Data governance + bias testing across patient cohorts
- +Technical documentation, logging, and traceability
- +Transparency to patients that AI is involved
Questions, answered
Is medical AI high-risk under the EU AI Act?
Yes. AI used for medical diagnosis, triage, or as a safety component of a medical device is classified high-risk under the EU AI Act, triggering conformity, oversight, and documentation duties.
Does the EU AI Act apply on top of MDR/IVDR?
Yes. The AI Act applies alongside the Medical Device and In-Vitro Diagnostic Regulations; conformity work can be aligned but the AI-specific obligations are additional.
When do healthcare AI obligations apply?
Transparency duties apply from 2 August 2026; the full high-risk obligations phase in by December 2027 under the Digital Omnibus.
Ask the Sovereign — the EU AI Act for healthcare
Governed answer · AI governance & cybersecurity only · signed to Layer 0
Open the full AI OS →