EU AI Act Article 6

The two routes to high-risk

The Annex I product route (Article 6(1)): the system is, or is a safety component of, a product covered by an EU safety law listed in Annex I, such as toys, medical devices, machinery or cars. That product must need a third-party conformity assessment because of health and safety risks. One required solely for other risks, such as radio-spectrum interference that does not affect health and safety, does not count (Article 6(1c)). A system used only for non-safety tasks, such as quality control, is not a safety component unless its failure would endanger health and safety (Article 6(1a) and (1b)).

The Annex III use-case route (Article 6(2)): the system is one of those named in Annex III, across eight areas from biometrics and employment to law enforcement and justice. Only the named systems count, such as one that screens job applications or scores creditworthiness.

The Annex III exemption

Article 6(3) lets an Annex III system escape high-risk status when it poses no significant risk of harm to health, safety or fundamental rights, and only if it meets one of four conditions, such as performing a narrow procedural task. An Annex III system that profiles natural persons is always high-risk.

Under Article 6(4), a provider using the exemption must document its assessment before placing the system on the market or putting it into service, register it in the EU database, and show the documentation to national authorities on request.

Why it matters

Article 6 is the gate to the Act’s high-risk chapter. The systems it classifies must meet requirements such as risk management and Article 15’s accuracy, robustness and cybersecurity rules, except in Section B products. A misclassified system either escapes duties that apply to it or takes on duties that do not.

The four exemption conditions

Under Article 6(3), an Annex III system is not high-risk if it poses no significant risk of harm, including by not materially influencing the outcome of a decision, and it is intended to:

• perform a narrow procedural task;
• improve the result of a previously completed human activity;
• detect patterns, or departures from patterns, in past decisions, without being meant to replace or influence the earlier human assessment without proper human review;
• perform a preparatory task for an assessment covered by Annex III.

Annex I has two sections

Annex I splits the product laws into two sections, and the section decides what the product route triggers.

Section A covers laws such as those for toys, lifts and medical devices. A system classified through Section A carries the Act’s high-risk requirements. Article 2(13) allows some of those requirements, in Articles 9 to 15 and 17 to 25, to be limited where the product law already protects health, safety or fundamental rights at least as well and overall protection does not drop. The Commission must say which systems and requirements this covers by 2 August 2027.

Section B covers other sector laws, including those for civil aviation security, motor and agricultural vehicles, marine equipment and unmanned aircraft. A system classified through Section B is still high-risk, but under Article 2(2) only Article 6(1), Article 60a and Articles 102 to 112 of the Act apply to it, plus Articles 57 to 59 in so far as the product law has absorbed the high-risk requirements. Articles 9 to 15 do not apply directly.

Machinery changed sections. The Digital Omnibus on AI deleted the old Machinery Directive 2006/42/EC from Section A and added the Machinery Regulation (EU) 2023/1230 to Section B, so an AI safety component in machinery now falls under the Section B rule.

When the rules start

The Digital Omnibus on AI, in force since 27 July 2026, moved the start dates. Sections 1 to 3 of the high-risk chapter, apart from Article 6(5), apply from 2 December 2027 for systems classified through Annex III, and from 2 August 2028 for systems classified through Annex I.

Questions and answers

How does the EU AI Act decide that an AI system is high-risk?

Through one of two routes in Article 6. The product route covers a system that is, or is a safety component of, a product under the EU safety laws listed in Annex I, where that product must pass a third-party conformity assessment for health and safety reasons. The use-case route covers the specific systems listed in Annex III, in areas such as biometrics, critical infrastructure, education, employment, access to essential services, law enforcement, migration, and justice and democratic processes.

When can an Annex III AI system avoid high-risk status?

Under Article 6(3), when it poses no significant risk of harm to health, safety or fundamental rights and is intended to do one of four things: a narrow procedural task, improving the result of a completed human activity, spotting patterns in past decisions without replacing or influencing the human assessment without proper review, or a preparatory task. A system that profiles natural persons is always high-risk. A provider relying on the exemption must document its assessment and register the system in the EU database under Article 49(2).

When do the Article 6 high-risk rules start to apply?

The Digital Omnibus on AI (Regulation (EU) 2026/1744) moved the dates. Sections 1 to 3 of the Act's high-risk chapter, apart from Article 6(5), apply from 2 December 2027 for systems classified under Article 6(2) and Annex III, and from 2 August 2028 for those classified under Article 6(1) and Annex I.