← All guides

Foundations

High-risk AI systems: classification and obligations

How a system becomes high-risk under Articles 6 and 7, the eight Annex III areas, the Article 6(3) filter, and the core duties that follow for providers and deployers.

Updated 3 July 2026 · Reference material, not legal advice

---

Most of the Regulation’s substance concerns high-risk AI systems. A system reaches that classification by one of two routes.

Two routes to high-risk (Article 6)

• Annex I route (Art. 6(1)). The system is a safety component of a product — or is itself a product — covered by Union harmonisation legislation listed in Annex I (machinery, toys, lifts, medical devices, vehicles, and so on), and is subject to third-party conformity assessment.
• Annex III route (Art. 6(2)). The system falls within one of eight sensitive use-case areas.

The eight Annex III areas

1. Biometrics — remote identification, biometric categorisation, emotion recognition.
2. Critical infrastructure — safety components of digital infrastructure, road traffic, and the supply of water, gas, heating, electricity.
3. Education and vocational training — admissions, evaluation of learning outcomes, exam-conduct monitoring.
4. Employment and worker management — recruitment and selection, task allocation, performance monitoring, promotion and termination.
5. Access to essential services — creditworthiness and credit scoring, life and health insurance pricing, public-benefit eligibility, emergency-call triage.
6. Law enforcement — victimisation risk, polygraphs, evidence-reliability assessment, offending-risk profiling.
7. Migration, asylum and border control — polygraphs, risk assessments, examination of applications.
8. Administration of justice and democratic processes — assisting judicial interpretation, influencing elections or voting behaviour.

The Article 6(3) filter

An Annex III system is not high-risk if it does not pose a significant risk of harm — for example because it performs a narrow procedural task, improves the result of a completed human activity, detects decision patterns without replacing human judgement, or performs a preparatory task. However, a system that carries out profiling of natural persons is always high-risk. A provider relying on this filter must document the assessment (Art. 6(4)) and register the system in the EU database (Art. 49(2)).

The core provider obligations (Articles 9–15)

Providers of high-risk systems must satisfy seven essential requirements: a continuous risk-management system (Art. 9); data governance with bias examination (Art. 10); technical documentation (Art. 11, Annex IV); automatic record-keeping / logging (Art. 12); transparency and instructions for use (Art. 13); designed-in human oversight (Art. 14); and appropriate accuracy, robustness and cybersecurity (Art. 15). They also operate a quality-management system, undergo conformity assessment, affix the CE marking and register the system.

What deployers must do (Article 26)

Deployers use the system per its instructions, assign competent human oversight, keep logs for at least six months, inform affected workers, and — for public bodies and some others — complete a fundamental-rights impact assessment (Article 27). Affected people also gain a right to an explanation of decisions (Article 86).

What this means in practice

High-risk classification turns a broad regulation into a specific, evidenced programme of work. Euridium maps each of these duties to its source article, scopes it to your role, and helps you assemble the evidence an auditor would ask for.

This guide is a plain-language summary for orientation. The authoritative text is Regulation (EU) 2024/1689 as published in the Official Journal of the European Union (12 July 2024). It does not constitute legal advice.