Annex III point 4
High-risk AI in HR and employment
For HR, recruitment, people analytics, platforms and organisations using AI in workforce processes.
Could the AI affect a person’s job opportunity, evaluation or working conditions?
EU AI Act · Annex III · 2026
High-risk AI guidelines for Annex III
Use these guides to determine faster whether an AI system may fall under the high-risk route, which questions to ask first and which evidence to collect.
High-impact Annex III areas
Annex III point 5
High-risk AI in essential services
For public bodies, banks, insurers, credit providers, benefit processes and emergency response.
Could the AI make access to an essential service easier, harder or more expensive?
Annex III point 3
High-risk AI in education
For schools, training providers, exam bodies, EdTech vendors and learning teams.
Does the AI support learning, or does it help decide something about a learner?
Annex III point 1
High-risk AI in biometrics
For organisations using biometric identification, biometric categorisation or emotion recognition.
Can this biometric AI system be used at all, or does a prohibition need to be checked first?
Also available
Annex III point 2
Critical infrastructure
For operators, suppliers and managers of digital infrastructure, traffic systems and utilities.
View guideAnnex III point 6
Law enforcement
For public bodies, suppliers and teams supporting investigation, evidence analysis or risk assessment.
View guideAnnex III point 7
Migration and border control
For public authorities, border operations, migration services and suppliers supporting asylum or border processes.
View guideAnnex III point 8
Justice and democracy
For legal organisations, public bodies, civic-tech vendors and teams supporting democratic processes.
View guideSource basis
These guides are based on the AI Act text and the European Commission draft classification guidelines of 19 May 2026. Check final guidance before completing legal decisions.
Want to know which route applies to your organisation?
The Gap Intake gives fast context: systems, role split, area, urgency and first evidence questions.