EUDAMED · MEDICAL DEVICES · EU
Where AI is named explicitly in the official device corpus
The monitor scans the public EUDAMED software corpus for exact AI text evidence. Manufacturer, risk, version and registry status remain separate observations—without deriving safety or conformity.
- official production API
- exact text evidence per candidate
- no AI inference from “software”
- no contact or PRRC data
Complete software corpus reviewedObserved since 23 July 2026
THE DECISIVE BOUNDARY
A text match is not evidence of effectiveness or conformity.
An exact term shows only what the public device record says. Risk class, status, manufacturer and version describe that same registry entry, but no causal effect.
text evidence ≠ technical assessment · registry status ≠ conformity verdictINTERACTIVE EVIDENCE TOPOLOGY
One device at the centre. Its published attributes around it.
Choose a candidate. The topology arranges fields from the same EUDAMED record—it claims neither causality nor certification.
Choose device
On small screens, scroll horizontally to explore the topology.
Does every software record in EUDAMED count as AI?
No. The monitor publishes a candidate only when an allowlisted public device field contains an exact AI term. Software classification, nomenclature, risk class and manufacturer name are never used as AI evidence.
Does the monitor show certificate changes?
No. The current official public API exposes actor, reference and UDI/device operations, but no certificate or market-surveillance operation. The monitor leaves those lanes absent instead of inventing a link.
Are first and last observed dates source change dates?
No. They are collector observation times. EUDAMED does not publish a source change timestamp in the records used here.
Does the monitor judge safety or conformity?
No. A registry status, risk class or text match is not a safety, clinical-performance or conformity assessment.