EU AI Act Risk Classification Guide
How to think about prohibited, high-risk, limited-risk, and minimal-risk AI systems under the EU AI Act.
Prohibited comes first
Potential Article 5 issues should be treated as stop-use or redesign triggers until reviewed. These include harmful manipulation, exploitation of vulnerable people, impermissible social scoring, and certain biometric uses.
High-risk is context-specific
A system can become high-risk because it is a safety component of a regulated product or because it is used in Annex III contexts such as employment, education, essential services, biometrics, law enforcement, migration, justice, or democratic processes.
Limited risk still needs evidence
Chatbots, AI interaction notices, deepfakes, synthetic content, and other transparency scenarios need clear disclosures and implementation evidence even when full high-risk duties do not apply.
Related EU AI Act guides
EU AI Act Compliance Checklist for SaaS Teams
A practical EU AI Act checklist for scope, risk classification, Article 50 transparency, and high-risk evidence.
EU AI Act Technical Documentation Template
What an EU AI Act technical documentation file should usually include for review and readiness.
EU AI Act Article 50 Transparency Requirements
A plain-English overview of Article 50 AI transparency duties and user disclosure evidence.