Skip to main content
Training & AI Literacy
6 min readUpdated 2026-02-15

AI Literacy Requirements (Article 4)

Understanding the Article 4 AI literacy obligation — who needs training, what it must cover, and how to demonstrate compliance through Klarvo's training tracking.

AI Literacy Requirements (Article 4)

Article 4 of the EU AI Act requires providers and deployers to take measures ensuring sufficient AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf. This obligation has applied since 2 February 2025.

What the Law Requires

Article 4 states that measures must:

  • Ensure sufficient AI literacy of staff
  • Take into account the technical knowledge, experience, education, and training of the persons involved
  • Consider the context in which the AI systems are to be used
  • Account for the persons or groups of persons on whom the AI systems are to be used
  • This means training should be role-appropriate — not one-size-fits-all.

    Who Needs Training

    Role CategoryTraining NeedPriority
    All staff using AI toolsBasic AI literacy: what AI is, how it works, limitations, responsible useHigh (applies since Feb 2025)
    AI system operatorsSystem-specific training: how to use the system correctly, interpret outputs, escalate issuesHigh
    Human oversight personnelAdvanced: system capabilities, failure modes, intervention procedures, authority to stopCritical (for high-risk)
    Reviewers/approversAssessment competence: how to evaluate classifications, evidence, and compliance artifactsMedium
    Leadership/boardGovernance awareness: AI risks, regulatory landscape, organisational responsibilitiesMedium

    What Training Must Cover

    Basic AI Literacy (All Staff)

  • What AI is and how it differs from traditional software
  • Common AI capabilities and limitations
  • Responsible AI use principles
  • Your organisation's AI acceptable use policy
  • How to report concerns about AI behaviour
  • System-Specific Training (Operators)

  • How the specific AI system works
  • What outputs it produces and how to interpret them
  • Known limitations and failure modes
  • When and how to escalate issues
  • Data handling and privacy requirements
  • Oversight Training (High-Risk System Oversight)

  • Deep understanding of the system's functioning
  • How to monitor for bias, drift, and errors
  • Authority and procedures for intervention/suspension
  • Incident response procedures
  • Regulatory obligations specific to high-risk deployers
  • Demonstrating Compliance

    To prove Article 4 compliance, you need evidence of:

  • Training program exists: Documented training plans by role
  • Content is appropriate: Training materials cover relevant topics
  • Training is delivered: Records of sessions conducted
  • Completion is tracked: Who completed what, when
  • Refresh is scheduled: Annual or more frequent re-certification
  • Effectiveness is assessed: Quiz results or competency checks (optional but recommended)
  • Using Klarvo for Training Tracking

    Klarvo's Training module lets you:

  • Define training programs by role category
  • Assign training to team members based on their AI system responsibilities
  • Track completion with dates and evidence
  • Set refresh cadence (typically annual)
  • Generate training reports for auditors — including a branded Training Compliance PDF Report showing completion rates, expiring training, and Article 4 compliance status
  • Link training to AI systems — showing which system operators have been trained
  • View the Training Gap Matrix — switch to Matrix view to see training gaps across your team at a glance, with a visual grid of team members vs training types
  • Training Content Sources

    Klarvo doesn't provide training content, but you can use:

  • Internal training materials (upload PDFs, link to videos)
  • External courses (MOOC platforms, vendor training)
  • Vendor-provided system training
  • Industry body resources
  • Klarvo's own documentation (this docs section covers many concepts)
  • Compliance Timeline

    AI literacy obligations apply since 2 February 2025. This means:

  • You should already have a training program in place
  • Completion records should be current
  • New staff should receive training before operating AI systems
  • Best Practices

    🎓 Role-based training: Don't give everyone the same training — tailor to responsibilities
    📋 Document everything: Training plans, materials, completion records, and quiz results are all evidence
    🔄 Annual refresh: AI evolves fast — retrain annually at minimum
    🏷️ Link to systems: Connect training records to the AI systems people operate
    Train before access: Ideally, no one should operate an AI system without completing relevant training