AI systems
AI System classification
Classification records the system's EU AI Act risk tier and the reasoning behind it. It sets the regulatory context; the Governance plan then shows the work that follows from it.
- In the app
- AI System → EU AI Act Classification (or the Classification tab)
Classification workflow
- 1Confirm the Scope Check first, or give an override reason.
- 2Open EU AI Act Classification and answer each step: Prohibited uses (Article 5), High-risk domains (Annex III) and Transparency (Article 50).
- 3Choose Classify system and review the result, the decision path and the articles and obligations it relies on.
Risk tiers
- Prohibited: the practice is banned under Article 5. Use Record retirement decision.
- High risk: the full set of EU AI Act obligations applies.
- Limited risk: transparency obligations apply.
- Minimal risk: no tier-specific obligations; general duties such as AI literacy still apply.
What happens on save
- Each run is saved to the system's classification history.
- If the project includes the EU AI Act framework, tasks tagged [EU AI Act] are created for the applicable obligations. Find them under Tasks.
- If the tier changes, active decisions for the system move to Review Required.
- When you edit scope facts later, the classification is marked out of date and the system prompts you to reclassify.
Last reviewed 2 October 2026
Next stepAI System Governance planThe Governance plan shows, for one AI System, which frameworks apply, which workspaces are required, and the controls, open work and risks that follow.