01
What Article 4 asks organisations to consider
The AI Act text addresses providers and deployers and asks them to take measures, to their best extent, to ensure a sufficient level of AI literacy for staff and other people operating or using AI systems on their behalf.
The measure is contextual. A board member, procurement professional, HR user, model developer and customer-support agent do not need identical depth, but each needs the competence required by their decisions and exposure.
02
Build role-based learning around real use
A practical programme starts from the AI inventory and identifies audiences, tasks, risks and decision rights. Training can then cover prohibited practices, escalation, human oversight, data handling, transparency and incident reporting at the appropriate depth.
Examples, exercises and assessments should use the organisation’s actual systems and policies. This is more effective than generic definitions and makes evidence easier to retain.
- Baseline awareness for all relevant personnel
- Role-specific modules for higher-impact decisions
- System-specific instructions and escalation routes
- Periodic refreshers triggered by change
03
Document measures and effectiveness
Keep the rationale for the programme, audience mapping, materials, attendance and review dates. Where appropriate, record assessments, feedback and updates after incidents or material changes.
Training evidence should demonstrate a considered programme, while avoiding the false claim that attendance alone proves compliance or competent use.