Skip to main content

Article 6: High-Risk AI System Classification

How Article 6 makes an AI system high-risk: the Annex I product route and the eight Annex III areas.

Providers use Article 6 to decide whether an AI system on the EU market or in service is high-risk.

Route 1: the system is, or is a safety component of, an Annex I product needing third-party conformity assessment.

Route 2: the system falls into an Annex III area, unless Article 6(3) applies.

The high-risk rules (Chapter III, Sections 1 to 3) apply from 2 December 2027 (Annex III) and 2 August 2028 (Annex I).

Misclassifying an Annex III system to avoid the rules can lead to fines (Article 80(7)).

Compare your system with Annex III and the Annex I laws.

Two routes make an AI system high-risk

An AI system is high-risk if it meets Route 1 or Route 2, and the two routes start on different dates.

Route 1: a product under an Annex I law, or its safety component

Annex I lists EU product-safety laws, for example for toys, lifts and medical devices. The product must also need a third-party conformity assessment (a check by an outside body).

For Section A laws, the high-risk rules (Chapter III, Sections 1 to 3) apply from 2 August 2028. For Section B laws, including machinery, only Articles 6(1), 60a and 102 to 112 apply (Article 2(2)). Articles 57 to 59 apply only where the sector law has integrated the high-risk requirements.

Read the Annex I guide: laws, dates and duties

Route 2: one of the eight Annex III areas

A system is high-risk if it falls into one of the eight areas listed in Annex III, such as recruitment or credit scoring.

The high-risk rules (Chapter III, Sections 1 to 3) apply from 2 December 2027. Four exemptions in Article 6(3) can apply, but profiling of people is never exempt.

Annex III names eight areas

Each area lists specific uses. A system built for one of those uses is high-risk, unless an exemption applies.

1
Biometrics
2
Critical infrastructure
3
Education and vocational training
4
Employment, workers’ management and access to self-employment
5
Access to and enjoyment of essential private services and essential public services and benefits
6
Law enforcement
7
Migration, asylum and border control management
8
Administration of justice and democratic processes

Read the Annex III guide: all eight areas with examples

Four conditions can exempt an Annex III system, but profiling never can

Article 6(3) says an Annex III system is not high-risk if it poses no significant risk of harm to the health, safety or fundamental rights of people. This applies where any one of these conditions is met:

  • The system performs a narrow procedural task.
  • The system improves the result of a human activity that was already completed.
  • The system detects decision-making patterns or deviations from earlier patterns. It is not meant to replace or influence the earlier human assessment without proper human review.
  • The system performs a preparatory task for an assessment that is relevant to the Annex III areas.

Profiling of people is never exempt: such a system is always high-risk (Article 6(3)).

A provider that relies on an exemption must document its assessment before the system is placed on the market or put into service, and must register the system (Article 6(4), Article 49(2)).

Scanara supports each classification step, and you make the decision

Find AI usage in your code

Scanara scans code for standard ML framework patterns. This supports finding which systems need a classification.

Record the classification

Scanara structures each classification and records your reasoning, including for an Article 6(3) exemption.

Keep the evidence

Scanara prepares classification documentation with an evidence trail. You decide the classification, with counsel where the stakes warrant it.

See Scanara’s features

Common questions about classifying AI systems

What if I classify my AI system wrongly?

If an Annex III system is misclassified as not high-risk to avoid the rules, the provider is subject to fines under Article 99 (Article 80(7)). Article 80(7) does not name an amount. Fines for breaking the provider duties in Article 16 reach EUR 15 million or 3% of worldwide annual turnover, whichever is higher (Article 99(4)). For SMEs and small mid-cap enterprises, the lower of the two amounts is the limit (Article 99(6) and (6a)). If you are unsure, document your reasoning and ask counsel.

Can an AI system become high-risk later?

Yes. If someone changes the intended purpose of a system that was not high-risk and is already on the market or in service, so that it becomes high-risk, that party becomes the provider (Article 25(1)(c)). An example is a chatbot repurposed for credit scoring. Reassess the classification whenever the purpose or context of use changes.

Do Article 6(3) exemptions remove all obligations?

No. The provider must document the assessment before the system is placed on the market or put into service and must register the system (Article 6(4), Article 49(2)). The exemption never applies to profiling of people (Article 6(3)). Other rules, such as the transparency duties in Article 50, can still apply.

When do the Annex I and Annex III rules apply?

The high-risk rules (Chapter III, Sections 1 to 3, except Article 6(5)) apply to Annex III systems from 2 December 2027 and to Annex I systems from 2 August 2028 (Article 113). Systems placed on the market or put into service before the relevant date are covered only after significant design changes; in any case, providers and deployers of high-risk systems intended to be used by public authorities must take the necessary steps to comply by 2 August 2030 (Article 111(2)). Article 113 gives each route its own date and does not say which date governs a system that falls under both routes. Check which route applies to your system.

After classifying: read your route’s guide, then the provider duties