AI literacy: what the AI Act asks of your team (and what it doesn't)

Does your team use ChatGPT for emails, Copilot in Word or an AI feature in your accounting software? Then your organisation falls under Article 4 of the EU AI Act: the rule on AI literacy. It has applied since 2 February 2025, but many business owners don't know exactly what is expected of them. This article sets it out.

What does Article 4 say?

In short: organisations that provide or use AI systems must take measures so their staff have sufficient knowledge of AI. The knowledge has to fit their work: someone who uses AI to write quotes needs to know how to check the output and which information must never go into an AI tool.

The rule doesn't only apply to tech companies. A ten-person installation company that uses ChatGPT is also a "deployer" in the sense of the AI Act.

What changed with the AI Omnibus?

In 2026 the AI Act was amended on several points through the so-called Digital Omnibus. For AI literacy it was clarified that this is an obligation of effort. You don't have to guarantee that every employee reaches a specific level, and regulators cannot require a specific course or certificate.

What remains: you must demonstrably work on AI knowledge in your organisation. Doing nothing is not an option, but you do have room to fill it in your own way.

What does "sufficient knowledge" mean in practice?

For most small and mid-sized businesses it comes down to four things every employee who uses AI should know:

1. What AI can and can't do. AI tools write convincing text, but they also make mistakes and sometimes invent facts.
2. How to check the output. Always verify figures, names, dates and sources before anything goes out.
3. Which information must not go into an AI tool. Customers' personal data and confidential business information don't belong in a free or personal account.
4. When to be open about AI. Since 2 August 2026 people must know when they are talking to an AI chatbot or seeing AI-generated content.

Four steps to sort it out

1. Map which AI you use

Make a simple overview: which AI tools do we use, who uses them, for what, and with which data? Don't forget AI features built into existing software, such as your CRM, accounting package or email client.

2. Make agreements

Write down on one page which tools are allowed, which information must not go in and how to check the output. Keep it short and clear; nobody reads a twenty-page document.

3. Train your team

A practical half-day workshop is enough for most teams to lay the foundation. It works best when employees practise with their own work: their own emails, quotes and reports. That way they learn not only the rules but also how to genuinely save time.

4. Record it

Keep track of who completed which training and when, and update your agreements when new tools are added. That is your evidence that you are working on AI literacy.

Common mistakes

• Thinking it doesn't apply to you. If anyone in your business uses AI for their work, Article 4 applies.
• Only writing a policy document. Rules without explanation aren't followed. Combine agreements with training.
• Banning AI. Bans rarely work: employees carry on by themselves, without any agreements. Controlled use is better.

Need help?

With our AI policy & AI Act package you sort out the overview, the agreements and the risk check in one week, at a fixed price. And with an AI workshop for your team you build the knowledge, including a certificate of attendance. Rather get started yourself? Download the free AI prompt guide, with 7 rules for working safely and a template for your AI usage rules.

This article is not legal advice. For complex situations, for example if you develop AI systems that fall under the high-risk rules yourself, we recommend a specialised lawyer.