Pricing
    About UsCareersContactPartnersPressStatus
    Start FreeLog in
    Classification & Risk Assessment
    10 min readUpdated 2025-01-28

    High-Risk Categories (Annex III)

    Complete reference guide to the nine high-risk AI use case categories defined in Annex III of the EU AI Act.

    High-Risk Categories (Annex III)

    Annex III of the EU AI Act lists specific use cases where AI systems are considered "high-risk" and subject to extensive compliance requirements.

    Category 1: Biometrics

    Scope: AI systems used for biometric identification, categorisation, or verification.

    Covered uses:

  1. Remote biometric identification
  2. Biometric categorisation (not prohibited)
  3. Real-time biometric verification
  4. Emotion recognition (beyond workplace/education)

    5. Examples:

  5. Airport facial recognition gates
  6. Age verification systems
  7. Access control biometrics
  8. Customer identification in banking

    9. Category 2: Critical Infrastructure

    Scope: AI as safety components of critical infrastructure management.

    Covered uses:

  9. Energy grid management
  10. Water supply control
  11. Digital infrastructure
  12. Road traffic management

    13. Examples:

  13. Smart grid load balancing
  14. SCADA system AI components
  15. Network traffic optimization
  16. Traffic light AI control

    17. Category 3: Education & Vocational Training

    Scope: AI affecting educational access, assessment, or opportunities.

    Covered uses:

  17. Admissions decisions
  18. Student assessment/grading
  19. Learning outcome evaluation
  20. Proctoring and cheating detection
  21. Educational resource access

    22. Examples:

  22. University admission scoring
  23. Automated essay grading
  24. Exam proctoring systems
  25. Learning management AI

    26. Category 4: Employment & Worker Management

    Scope: AI systems affecting employment decisions and worker management.

    Covered uses:

  26. Recruitment and screening
  27. Job advertising targeting
  28. CV/application filtering
  29. Interview assessment
  30. Performance evaluation
  31. Task allocation
  32. Promotion/termination decisions
  33. Workplace monitoring

    34. Examples:

  34. ATS candidate ranking
  35. Video interview analysis
  36. Performance management AI
  37. Shift scheduling systems
  38. Productivity monitoring

    39. Category 5: Essential Private & Public Services

    Scope: AI affecting access to essential services and benefits.

    Covered uses:

  39. Creditworthiness assessment
  40. Credit scoring
  41. Risk assessment for insurance (life/health)
  42. Emergency service dispatch
  43. Public benefit eligibility
  44. Social service allocation
  45. Healthcare access prioritization

    46. Examples:

  46. Bank loan decisioning
  47. Insurance premium calculation
  48. Emergency 911/112 triage
  49. Benefit fraud detection
  50. Hospital bed allocation

    51. Category 6: Law Enforcement

    Scope: AI supporting law enforcement activities.

    Covered uses:

  51. Evidence reliability assessment
  52. Crime risk assessment (polygraph, etc.)
  53. Profiling during investigation
  54. Lie detection
  55. Crime pattern prediction
  56. Deep fake detection in evidence

    57. Examples:

  57. Forensic evidence analysis
  58. Risk assessment tools for courts
  59. Investigative AI assistants
  60. Predictive analytics (location-based)

    61. Category 7: Migration, Asylum & Border Control

    Scope: AI in immigration and border management contexts.

    Covered uses:

  61. Polygraph/similar tools
  62. Security/health/environmental risk assessment
  63. Visa/asylum application processing
  64. Document verification
  65. Petition/complaint examination
  66. Irregular migration risk assessment

    67. Examples:

  67. Visa decision support
  68. Asylum case prioritization
  69. Border crossing risk assessment
  70. Document fraud detection

    71. Category 8: Administration of Justice & Democratic Processes

    Scope: AI assisting judicial and democratic institutions.

    Covered uses:

  71. Researching legal facts
  72. Applying law to facts
  73. Alternative dispute resolution
  74. Election result influence (advertising excepted)
  75. Voting behavior influence

    76. Examples:

  76. Legal research AI
  77. Sentencing recommendation systems
  78. Contract analysis tools
  79. Voter registration assistance

    80. Category 9: Safety Components of Regulated Products

    Scope: AI that is a safety component of products covered by EU product legislation (Annex I).

    Covered sectors:

  80. Medical devices
  81. Motor vehicles
  82. Aviation
  83. Marine equipment
  84. Toys
  85. Machinery
  86. Lifts
  87. Personal protective equipment

    88. Examples:

  88. ADAS/autonomous driving
  89. Medical diagnostic AI
  90. Industrial robot safety
  91. Drone flight control

    92. Implications of High-Risk Classification

    If your AI system matches any Annex III category, you become subject to:

    For Deployers (Article 26):

  92. Use according to provider instructions
  93. Assign competent human oversight
  94. Ensure input data quality
  95. Monitor operation
  96. Keep logs ≥ 6 months
  97. Report incidents
  98. Conduct FRIA (if applicable)

    99. For Providers (Chapter III, Section 2):

  99. Risk management system
  100. Data governance
  101. Technical documentation
  102. Logging capability
  103. Transparency & user info
  104. Human oversight design
  105. Accuracy & robustness
  106. Conformity assessment
  107. Registration in EU database