AI Literacy: What Article 4 of the AI Act Requires
AI Act Article 4 requires businesses that use AI to make their staff AI-literate. Already in force, enforceable from 2 August 2026. What to do.
Gepubliceerd door NordX Consulting — AI bureau voor enterprise bedrijven in Nederland.
AI literacy sounds like a soft recommendation. It is a legal duty. Article 4 of the EU AI Act requires every business that deploys AI to make sure the people working with that AI understand it well enough. This obligation has applied since 2 February 2025, and from 2 August 2026 supervisory authorities can enforce it.
What Article 4 actually says
The text is short. Providers and deployers of AI systems must take measures to ensure, to their best extent, a sufficient level of AI literacy among their staff and other persons operating and using AI systems on their behalf.
Two things stand out. First: it applies not only to builders of AI (providers), but also to ordinary businesses that use AI (deployers). If your team uses ChatGPT, an AI chatbot, or an AI feature in your accounting software, you fall under Article 4. Second: it also covers people who operate AI "on your behalf" — contractors and vendors running AI for you.
AI literacy is defined in the law (Article 3(56)) as the skills, knowledge, and understanding that allow someone to deploy AI systems in an informed way and to become aware of the opportunities, the risks, and the possible harm.
Who does it apply to?
In short: nearly every business that uses AI. There is no exception for small businesses, no size threshold, and no limitation to "high-risk" systems. If you use AI in your work processes, the people working with it must be AI-literate — appropriate to their role.
That "appropriate to their role" matters. The law does not require a diploma or exam. A marketer using AI for copy needs different understanding than a developer training an AI model. The level has to match what someone does with AI and in what context.
What "literate" means in practice
The European Commission does not prescribe a mandatory curriculum, but the law and its guidance paint a clear picture. Your people should:
- Understand what the system does — its basic function, what it can and cannot do, and where the limits lie.
- Recognise the risks — hallucinations, bias, privacy, and when a human must check an output.
- Know when AI is not suitable — which decisions you do not leave to a model.
- Know the rules — how to handle company data and personal data when using AI.
So it is not about technical depth; it is about responsible use.
The link with 2 August 2026
Why now? Article 4 already applies, but the enforcement structure of the AI Act — the national supervisory authorities and the associated penalties — only becomes applicable on 2 August 2026. From that point there is a body that can inspect and act. It is the same date on which the transparency duty of Article 50 takes effect.
The AI Act does not attach its own fine tier to Article 4, the way it does to the prohibited practices or the transparency duty. But it remains a legal obligation, and in a broader compliance review, demonstrably AI-literate staff is the difference between "we handled it" and "we had no idea." For what was and was not deferred, see the delay analysis.
What to do now
1. Map who uses AI. Not only the obvious tools. AI buried inside existing software (CRM, accounting, customer service) counts too.
2. Decide what "sufficient" means per role. A team using AI for content needs different knowledge than a team building AI decisions into a process.
3. Provide basic training. It does not have to be an expensive course. An internal session on what your AI tools do, where the risks are, and what the house rules are covers the core for most teams.
4. Record that you did it. A short note — who was trained, on what, when — is your proof that you meet Article 4. Without documentation, "we discussed it" is hard to stand behind.
5. Make it repeatable. AI changes fast; a one-off session ages. Build AI literacy into onboarding and repeat it periodically.
In short
Article 4 requires every business that uses AI to make its people AI-literate, appropriate to their role. The duty has applied since February 2025 and becomes enforceable from 2 August 2026. It is not a heavy project: inventory who uses AI, arrange basic training, and record that you did it. The businesses that pick this up now will not be caught out later.
Sources
Frequently asked questions
Does the AI literacy duty also apply to small businesses?
Yes. Article 4 has no exception for small companies and no size threshold. Every business that uses AI — including through existing software — must ensure the staff involved are AI-literate, appropriate to their role.
Since when does Article 4 apply?
Article 4 has been in force since 2 February 2025. The enforcement structure of national supervisory authorities and penalties becomes applicable from 2 August 2026.
Do I have to send my staff on an official course?
No. The law does not prescribe a diploma or exam. What is required is a level of AI literacy appropriate to the role: understanding what the AI does, recognising the risks, and knowing the house rules. An internal training session is often enough.
What is the penalty for non-compliance with Article 4?
Article 4 does not carry its own fine tier like the prohibited practices or the transparency duty. It remains a legal obligation, and demonstrable AI literacy is part of a broader compliance file.
Meer weten? Bekijk onze andere artikelen op het NordX blog of neem contact op via nordx.ai.