What is AI literacy under the EU AI Act?

AI literacy is an organisational measure, not a one-off course

Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures, to their best extent, to ensure a sufficient level of AI literacy among staff and others dealing with AI systems on their behalf. The appropriate level depends on matters including technical knowledge, experience, education, training and the context in which AI is used.

The requirement has applied since 2 February 2025. The Commission's AI literacy FAQ makes clear that there is no single mandatory training format or certificate specified by Article 4.

What staff should understand

• Which AI systems and features they are permitted to use for work.
• The limits of AI outputs, including hallucination, bias and context errors.
• Which personal, confidential or sensitive information must not be entered without an approved basis and tool.
• When human verification is required before an output is relied upon or shared.
• How to escalate an error, unexpected behaviour or suspected incident.
• Any role-specific risks created by the systems they operate or oversee.

Make literacy role-appropriate

A general awareness session may be useful for the whole workforce, but it should not be the end of the programme. Procurement teams need to understand supplier questions. Managers approving AI use need risk and escalation criteria. People overseeing higher-impact systems need more detailed knowledge of limitations, monitoring and intervention.

What evidence should you keep?

Article 4 does not prescribe a specific certificate. As a practical governance matter, organisations can keep evidence of the training or guidance provided, the groups covered, relevant policy acknowledgements and how materials are updated when approved tools or risks change.