Somebody has asked whether you comply.
Article 4 of the EU AI Act requires providers and deployers of AI systems to ensure a sufficient level of AI literacy among the staff who use them. It has applied since February 2025, and national supervision began in August 2026.
If you operate in the EU, or sell to organisations that do, this has probably landed on your desk with a deadline attached.
What it requires, stated plainly
Your people need to understand the systems they use: what those systems can be trusted with, where they fail, and what stays a human decision. That is the substance of it.
What it does not do is carry direct fines. We will not tell you otherwise to sell you training, and you should be careful of anyone who does.
The trap is meeting it badly
The cheapest compliant answer is a video and a completion certificate. It satisfies an auditor and changes nothing about how anybody works, and you will pay for it again the first time someone pastes something they should not into a chat window.
The obligation is an opportunity to do the thing you would want to do anyway: give people working fluency, so the risk drops because judgement improved.
What we do
Literacy that changes behaviour. What AI is reliable for, how to give it what it needs, and where a human must stay accountable. It maps to Article 4, and it is worth doing whether or not the regulation existed.