AI literacy

What the law actually says

Article 4 of the EU AI Act requires providers and deployers to take measures ensuring, to their best extent, a sufficient level of AI literacy among their staff and other persons operating AI systems on their behalf. It has applied since 2 February 2025.

Two things are worth reading carefully. The obligation applies to deployers, not only to companies that build AI, so ordinary businesses using bought-in tools are in scope. And the standard is relative: it must take into account technical knowledge, experience, education and the context in which the systems are used.

What sufficient means in practice

The law does not prescribe a course or a certificate. It asks for a level appropriate to the role. A planner who lets a model propose a schedule needs to know when to distrust it. A director signing off on the system needs to understand what it can and cannot be held to. A warehouse employee using a chat assistant needs to know what not to paste into it.

One generic training for everyone satisfies the letter and misses the point.

Why this is a capability question

AI literacy is the part of AI Capability that the law happens to make compulsory. An organisation where only one person can judge an AI output has a compliance problem and a continuity problem at the same time.

A reasonable minimum

• Write down which AI systems are in use and who operates them
• Set the level per role rather than per person
• Cover judgement, not just operation: when is the output wrong and what do you do then
• Record what you did, because the obligation is on the measures you take