Risk classification
Classify the organization's AI systems using the EU AI Act four-tier risk framework. The classification form shows a rule screening of the system's registered facts (prohibited practices, Annex III areas, transparency duties) with a suggested level; you choose the level and, for high risk, the Annex III category. Every classification is kept in the history for audit purposes.
Risk tiers
The EU AI Act defines four risk levels. Each level carries different compliance obligations, from outright prohibition to minimal transparency requirements.
Unacceptable
AI systems that pose an unacceptable risk to safety or fundamental rights. These are prohibited under Art. 5.
Examples: Social scoring, real-time biometric identification in public spaces, manipulation techniques.
High
AI systems listed in Annex III or used as safety components. Subject to full compliance requirements (Art. 6-51).
Examples: Biometric identification, critical infrastructure, employment decisions, law enforcement.
Limited
AI systems with specific transparency obligations. Users must be informed they are interacting with AI (Art. 50).
Examples: Chatbots, deepfake generators, emotion recognition systems.
Minimal
AI systems posing minimal risk. No mandatory compliance requirements, though voluntary codes of conduct are encouraged.
Examples: Spam filters, AI-enabled video games, inventory management systems.
Annex III categories
The classification form lists the Annex III high-risk categories. If the AI system falls into any of these areas, it is likely classified as High Risk.
Classification history
Every classification change is versioned with a timestamp, the author, the previous and new risk levels, and a rationale. This provides a complete audit trail for regulators.
Initial classification based on chatbot use case.
Re-classified after scope change to include employment screening.
How to classify an AI system
AI officerOpen the AI system
Navigate to the AI Registry and select the system you want to classify.
AI officerOpen the classification form
Choose Classify for that system to open the risk classification form.
AI officerRead the rule screening
The form shows which rules matched the system's registered facts (prohibited practices, Annex III areas, transparency duties) and the level they suggest. It is a triage aid, not a legal determination.
AI officerChoose the risk level
Choose the risk level, accepting the suggestion or not, and for high risk the Annex III category. The level you choose determines the compliance obligations.
AI officerDocument the rationale
Add a rationale explaining the classification decision. This is stored in the classification history for audit purposes.