Commission guidelines on classifying high-risk AI systems

Regulator guidanceHigh-risk guidelines (Art. 6)

Summary

They explain with examples when an AI system is high-risk and when it is not. The intended purpose, as described in product materials, is the key test. They clarify that candidate screening cannot use the Article 6(3) exception because it evaluates personal aspects.

Who it applies to

Providers and deployers who need to classify their AI systems.

Penalties

The rule has no penalty regime of its own, or it does not apply.

Official name: Draft Commission Guidelines on the classification of high-risk AI systems (Article 6 of Regulation (EU) 2024/1689)
Official reference: Borrador publicado por la Comisión Europea el 19 de mayo de 2026; versión final prevista para finales de 2026
Regulator: European Commission
Official source: Check the sourceHow we verify sources

Key dates

May 19, 2026Draft published
June 23, 2026Public consultation closes

Obligations and dates

Obligations table
Obligation	Article or citation	Status	Date	Who it applies to	Verification
Classify each AI system and document the reasoning	Reglamento (UE) 2024/1689, art. 6(3)-(5); borrador de directrices de 19-may-2026	Regulator guidance	Already applicable	If you use or sell AI in areas that could be high-risk, or if you combine several components or agents that influence a decision.	Verified in the official source on October 31, 2026

Classify each AI system and document the reasoning

Reglamento (UE) 2024/1689, art. 6(3)-(5); borrador de directrices de 19-may-2026Regulator guidance

Already applicable

When it relates: If you use or sell AI in areas that could be high-risk, or if you combine several components or agents that influence a decision.
What the rule provides: Classify each system by its intended purpose and keep the rationale.
Do not apply the Article 6(3) exception to systems that profile or evaluate people, such as CV screening.
Treat as a single system any set of components or agents whose combined output materially influences a decision.
Typical evidence: Classification matrix with per-system rationale.
Up-to-date AI system inventory.
Related base documents: AI Risk Classification Matrix
AI System Inventory
Supporting documents: AI Use Case Portfolio
AI Impact-Risk Prioritization Matrix
Verification: Verified in the official source on October 31, 2026Draft published on the Commission website (official source cited in the research). It is a draft and may change in the final version.