People and governance

Article 4 AI literacy: a practical plan

AI literacy is not satisfied by a single generic training slide. A proportionate programme should reflect the knowledge, experience and context of the people who provide or use AI systems.

Prepared by EU AI Fit editorial team · Published 24 August 2026 · Source review 31 August 2026

Segment the audience

Differentiate general users, system owners, developers, reviewers, procurement teams and senior decision-makers. Each group needs enough knowledge to perform its responsibilities safely.

Connect training to real systems

Cover intended use, limitations, data handling, human oversight, escalation and incident reporting for the systems people actually use. Higher-impact contexts justify deeper and more frequent learning.

Evidence the programme

Keep the measures taken, curriculum, attendance, role mapping, assessments, refresh dates and change history. Capture how learning needs were selected and how effectiveness is reviewed. The amended Article 4 does not require an organisation to guarantee a specific level of AI literacy for any individual.

Recommended next step

Map each inventoried system to the people who operate or govern it, then define the knowledge and evidence required for those roles.

Run the free exposure check

Related practical guides

Put the guidance into practice

Continue your EU AI Act review