Aurorisk FR

Chapter 1 · Lesson 1.3

Qualify a system under the AI Act

The European AI Regulation (“AI Act”) classifies systems by risk level. Qualification follows a precise order, and it decides the obligations that follow.

OrderQuestionConsequence
1General-purpose AI model?Obligations specific to model providers (Art. 51 to 55)
2Prohibited practice (Art. 5)?Prohibited: social scoring, manipulation, emotion recognition at work or school, etc.
3Safety component of an Annex I product, assessed by a third party?High risk (Art. 6(1))
4Use listed in Annex III?High risk (Art. 6(2)), except under the exception in Art. 6(3)
5Interaction with people, generated content, emotions?Transparency obligations (Art. 50)
—None of the previous casesMinimal risk: no specific obligation, except AI literacy (Art. 4)
The AI Act qualification orderFive questions in order: general-purpose model, prohibited practice, safety component of an Annex I product, Annex III use, interaction or generated content. Each “yes” leads to a regime: model provider obligations, prohibition, high risk, high risk subject to exception, transparency. If every answer is no, the risk is minimal.1. General-purposemodel?yesObligations of themodel provider (Art. 51 to 55)no2. Prohibited practice(Art. 5)?yesProhibitedno3. Annex I, assessedby a third party?yesHigh risk(Art. 6(1))no4. Annex IIIuse?yesHigh risk, exceptunder Art. 6(3)no5. Interaction, generatedcontent, emotions?yesTransparency(Art. 50)noMinimal riskQuestions 3 and 4lead to high risk;transparency maybe added: we judgethe use, not the nameof the tool.
Qualification follows this order. A “yes” gives the regime; high risk and transparency can add up, as for a chatbot that would also screen candidates.

Annex III in one line per area

Biometrics; critical infrastructure; education and training; employment and worker management (recruitment, selection, evaluation, decisions on working conditions); access to essential services (including credit scoring); law enforcement; migration and asylum; justice and democratic processes.

The exception in Art. 6(3)

An Annex III system is not high risk if it does not pose a significant risk to health, safety or fundamental rights, for example if it is limited to a narrow procedural task or prepares human work. Never if the system profiles people. A provider who invokes the exception must document its reasoning.

Key takeaways

Qualification follows an order: general-purpose model, prohibited practice, Annex I, Annex III, transparency. We judge the use, not the name of the tool; high risk can add up with transparency. For Annex III, high-risk rules apply from 2 December 2027.

Updated: November 2026. Application timetable, amended by Regulation (EU) 2026/1744 (“digital omnibus on AI”, in force since 27 July 2026): prohibited practices and AI literacy since 2 February 2025; general-purpose models since 2 August 2025; transparency (Art. 50) since 2 August 2026; Annex III high-risk systems from 2 December 2027, Annex I from 2 August 2028. Re-read the consolidated text on EUR-Lex at each review (lesson 5-3).