Annex III High-Risk AI Systems: Article 6 Classification Guide
Article 6 of the EU AI Act defines two pathways to high-risk classification. This guide walks you through both, the Annex III use-case categories, the Article 6(3) exemption, and a self-assessment flowchart.
Last updated: April 29, 2026
What Makes an AI System High-Risk?
Article 6 creates two independent pathways. Falling under either makes you high-risk — and triggers the full Annex IV documentation, Article 9–15 controls, and conformity assessment obligations.
The two pathways: Article 6(1) covers AI used as safety components in regulated products. Article 6(2) + Annex III covers stand-alone AI in eight specific use-case categories.
Pathway 1: Safety Components (Article 6(1))
If your AI is a safety component of a product covered by EU sectoral legislation (medical devices, vehicles, machinery, lifts, toys, etc.), or is itself such a product, it is automatically high-risk.
Conformity assessment must be conducted by a notified body.
Pathway 2: Annex III Use Cases (Article 6(2))
Eight use-case categories trigger high-risk classification regardless of the underlying technology:
- Biometrics: Remote biometric identification and emotion recognition systems
- Critical Infrastructure: AI in energy, water, gas, heating, and digital infrastructure
- Education: AI for student assessment, exam scoring, and learning analytics
- Employment: CV screening, interview analysis, task allocation, performance evaluation, termination decisions
- Essential Services: Credit scoring, loan approval, insurance pricing, emergency services dispatch
- Law Enforcement: Risk assessment, polygraphs, evidence evaluation, profiling
- Migration & Asylum: Visa applications, asylum processing, border control
- Justice: Sentencing assistance, case outcome prediction, legal research
The Article 6(3) Exemption
An Annex III AI system can be exempted if it does not pose a significant risk to health, safety, or fundamental rights.
Recognized cases include narrow procedural tasks, improving the result of a previously completed human activity, detecting decision patterns without replacing human assessment, or preparatory tasks for a downstream human decision. Profiling is always high-risk. The exemption never applies if the system performs profiling of natural persons.
If you claim the exemption, document the reasoning thoroughly. Regulators may challenge it.
Documentation Requirements If High-Risk
High-risk AI systems must maintain comprehensive compliance documentation:
- Annex IV technical documentation
- Article 9 risk management system
- Article 10 data governance records
- Article 12 logging in production
- Article 14 human oversight design
- Conformity assessment + CE marking
- EU AI database registration