AI cybersecurity threat model & evidence

Dated AI-specific threat model (data poisoning, model poisoning, adversarial examples, confidentiality attacks) with mitigations and validation.

Artifact type: threat-model
Record basis: Contemporaneous — create it when the decision is made
Annex IV section: anx-4-3
Primary source: Art. 15 AIA

Which obligations this evidences

Evidence AI-specific cybersecurity

Legal source

Art. 15 AIA

Stay audit-ready as the Act changes

EU AI Regulation Decoded tracks each obligation and the evidence auditors expect — one practitioner email a week.

Subscribe — free, weekly

Not legal advice. This is a practitioner reference generated from a cited knowledge base built on the primary legal text and official Commission guidance. Verify against the cited primary sources before relying on it for a compliance decision. See editorial standards & methodology.