AI literacy after the AI Act changes: what a business must do
Article 4 of the AI Act changed in July 2026. Businesses still have to support their team's AI literacy. What changed, and how to do it in 5 steps.


Contents
AI literacy means the skills, knowledge and understanding that let people use AI systems in an informed way and be aware of their opportunities, risks and possible harm. That’s how Article 3(56) of the AI Act defines it. Article 4 requires businesses that build or use AI to look after this literacy among their people.
The rule changed in July 2026. Below we explain exactly what changed, then walk through five steps to meet the duty without stress. Legal status as of 22 October 2026. This is not legal advice.
What did Article 4 say, and what does it say now?
The original Article 4 has applied since 2 February 2025. It said that providers and deployers of AI must take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and others operating AI systems on their behalf.
The Digital Omnibus, Regulation (EU) 2026/1744, changed that wording. It entered into force on 27 July 2026. Businesses now have to take measures to support the development of AI literacy. The new text states plainly that the duty does not require guaranteeing any specific level of AI literacy of any individual. On top of that, the European Commission and Member States are to support businesses here, with particular attention to SMEs.
Who does it apply to?
Every business that uses AI at work, even if it’s “just” ChatGPT for emails. The Article 4 duty doesn’t depend on the system’s risk level. It covers employees and others who operate AI systems on the company’s behalf, such as contractors.
Step 1: List the AI tools your business uses. Who uses them, and for what?
Start with a list: tool, who uses it, for which tasks. Without it you don’t know what to teach. In many companies it turns out there are more tools than the boss assumed, because some people use personal accounts.
Checklist:
- Which AI tools are in use, including on personal accounts?
- Who uses them, and for which tasks?
- Does any tool process customer data or personal data?
Step 2: Set simple rules. What’s allowed and what isn’t?
Rules are the cheapest way to raise literacy. One page: which tools we use, which data we don’t paste in (for example, customer personal data without a legal basis, or company secrets), who checks outputs before they go to a client, how we label AI content. You can see an example in our AI use policy.
Checklist:
- Does the team know which data must not be pasted into AI tools?
- Who checks AI outputs before they go out?
- Do the rules cover labelling AI content (Article 50 of the AI Act)?
Step 3: Train people by role. Who needs to know what?
The definition in Article 3(56) talks about technical knowledge, experience and the context of use. A salesperson writing proposals with AI needs different knowledge from an accountant pasting reports into it. One general session for everyone is a good start, but short modules for specific roles are worth adding.
We helped prepare 12 AI Trust CERT courses, including an AI Literacy course, and we run training on using AI safely for teams. What we’ve learned: examples from participants’ daily work beat generic slides.
Checklist:
- Has everyone who uses AI had at least basic training?
- Does the training cover the risks typical for each role?
- Do new hires get the training as part of onboarding?
Step 4: Keep a paper trail. How will you show the business is acting?
The new Article 4 is about measures, and measures are easy to document. Keep the list of tools, a dated version of the rules, attendance lists and training materials. That’s your evidence that the company takes the duty seriously.
Checklist:
- Do we have attendance lists and materials from every session?
- Do the rules have a date and a version number?
- Do we know where these documents are if someone asks?
Step 5: Come back to it. When should the team refresh its knowledge?
AI tools change every few months, and so do the rules. Set a fixed review point: for example, when tools change, after new guidance, or every six months. A quick update to the rules and 30 minutes with the team is enough.
Checklist:
- When did we last update the rules?
- Who in the company follows changes in tools and regulation?
- Does the team know who to ask when they’re unsure about AI?
On your own or with help?
You can do steps 1, 2 and 4 yourself in an afternoon. Role-based training is easier with someone who knows both the tools and the rules. If you want to do it once and do it properly, book a free call.
Written rules are a good first step, and you can follow our guide to a company AI policy.
Summary
Article 4 of the AI Act still applies. Since 27 July 2026, a business must support the development of AI literacy, without guaranteeing a set level for each person. Start with a list of tools and simple rules. Train people by role and keep the paperwork. Come back to it regularly, because tools and rules keep changing.
One thing to do today: write down, on one page, which data the team must not paste into AI tools. That’s already the kind of measure Article 4 talks about.
Sources
- Regulation (EU) 2024/1689 (AI Act), Article 3(56) and Article 4, EUR-Lex.
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal of the EU, 24 July 2026.
- AI literacy: Digital Omnibus rewrites Article 4 of the AI Act, Law and Technology.
- AI Act
- Article 4
- AI literacy
- training
Written with the help of AI, edited and checked by a human.



