The AI literacy duty has applied since 2 February 2025 and covers companies that develop AI systems or use them in their work. In July 2026 it was rewritten by the digital omnibus package on AI, Regulation (EU) 2026/1744.
Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), Regulation (EU) 2026/1744, digital omnibus on AI (in Czech)
What is AI literacy?
The AI Act defines it in Article 3(56) as the skills, knowledge and understanding that allow people to make informed use of AI and to be aware of its opportunities, risks and possible harm. People should know what an AI tool can do, where it gets things wrong and what to watch out for. They do not need to program or understand how the models work.
Source: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech)
What changed in Article 4?
The duty applies to providers of AI systems (those who develop AI and place it on the market) and to deployers (those who use AI in their work). In practice this includes companies whose staff use a chatbot or AI features in everyday software.
| Original wording (from 2 February 2025) | New wording (from 27 July 2026) | |
|---|---|---|
| What the company does | takes measures to ensure, to its best extent, a sufficient level of AI literacy | takes measures to support the improvement of AI literacy |
| Who it covers | staff and other persons dealing with the operation and use of AI on its behalf | unchanged |
| Level for individuals | a sufficient level | the regulation expressly does not require a specific level to be guaranteed |
| Support for companies | not mentioned | the Commission and Member States support small and medium-sized enterprises in particular; the Commission will publish practical examples |
The new wording has applied since Regulation 2026/1744 entered into force on 27 July 2026. According to its recitals, a strict obligation to ensure a sufficient level would not suit every type of company and would place an additional burden on smaller businesses in particular.
So the duty has not gone away. Its content has changed: instead of ensuring a sufficient level as far as possible, the company takes measures to support AI literacy. An internal record helps to show what it has done.
Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), Regulation (EU) 2026/1744, digital omnibus on AI (in Czech), European Commission, AI Act Service Desk: Article 4, Ministry of Industry and Trade, AI Act amendment and new guidelines (in Czech)
What does the European Commission recommend?
The Commission publishes questions and answers on AI literacy, most recently updated on 20 October 2026, after the change. Several practical points follow from them:
- Start with how you use AI. The company should have a general understanding of AI, know its role (provider or deployer) and know the risk of the systems it uses. It then adapts its measures to its people's knowledge and to the setting in which they use AI.
- No certificate is needed. The company can keep an internal record of training or other measures.
- The Commission does not prescribe a format. The regulation does not require mandatory training, and there is no need to appoint a dedicated "AI officer". The Commission does point out that relying on the instructions for use alone is often not enough. It does not require companies to test their employees' knowledge.
- Everyday tools count too. When staff use generative AI for texts or translations, they should know its risks, for example that the model can make things up.
- "Other persons" go beyond employees. They can include contractors or service providers who work with AI on the company's behalf.
The Commission also keeps a public collection of practical examples, the so-called living repository, with more than 40 initiatives from various organisations. It is meant as inspiration. Adopting a practice listed there does not in itself guarantee that a company has met the duty.
Sources: European Commission, AI Literacy: Questions & Answers, European Commission, Living repository on AI literacy
Is there a fine for breaching Article 4?
The regulation sets no fixed upper limit on fines for breaching Article 4. Article 99(4), which lists the obligations carrying fines of up to EUR 15 million or 3% of turnover, does not include Article 4. Penalties for any breach of the regulation are, however, set by Member States and may include fines, warnings and other measures.
According to the Commission, national market surveillance authorities have enforced the duty since 2 August 2026, and penalties should be proportionate. The Commission says enforcement may be more likely where an incident is shown to have been caused by insufficient staff training. As far as available sources show, the Czech Artificial Intelligence Act, which will designate supervisory authorities and penalties, had not been adopted as of 31 October 2026. A specific penalty in Czechia therefore cannot yet be stated.
High-risk systems are a separate matter. Under Article 26, deployers will have to assign human oversight to people with the necessary competence and training. This obligation sits alongside Article 4 and the amendment did not soften it. It only moved its date of application to 2 December 2027 (systems under Annex III) and 2 August 2028 (Annex I).
Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), European Commission, AI Literacy: Questions & Answers, Regulation (EU) 2026/1744, digital omnibus on AI (in Czech)
How do you meet the duty in practice?
For most companies a simple approach that can be documented is enough:
- List where you use AI. Which tools, who works with them and what for. This includes AI features inside office software, CRM systems or development tools.
- Group people by how they work with AI. Someone who occasionally has AI translate an email needs a different level from a developer building AI into a product, or an HR specialist who uses it to screen applicants.
- Choose proportionate measures. A short internal rule for working with AI, training on the risks (errors, confidential data, copyright), and a specialist course for more demanding roles.
- Keep a record. Who completed which measure and when. A certificate is not required, but a record is useful if the supervisory authority asks.
- Update regularly. New tools and new people change what staff need to know.
Summary
- The AI literacy duty under Article 4 of the AI Act still applies. Since 27 July 2026 companies take measures to support their staff's AI literacy but do not have to guarantee a specific level.
- The Commission requires no certificate, mandatory training or dedicated role. It makes sense to have an overview of how AI is used, proportionate measures and a record of them.
- The regulation sets no fixed fine for Article 4. Penalties will be set by Czech law, but the Czech AI act has not yet been adopted.
We cover the other AI Act obligations and their deadlines in the article AI Act in 2026: what companies should already have in place and what comes next and on the page AI Act: your company's readiness. If you are introducing AI in your company, write to us about the tools you are planning and who will use them. Our approach is described on the AI solutions page.
Status as of 31 October 2026. This article is general information, not legal advice.
- AI Act
- AI literacy
- training
Sources
- Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 3(56), Article 4, Article 99 (in Czech)
- Regulation (EU) 2026/1744, Article 1 points 5, 38 and 40 (new wording of Article 4, Article 99(1) and Article 113), recital 8, Official Journal 24 July 2026 (in Czech)
- European Commission: AI Literacy – Questions & Answers (updated 20 October 2026)
- European Commission: Living repository to foster learning and exchange on AI literacy (updated 10 August 2026)
- European Commission, AI Act Service Desk: Article 4
- Ministry of Industry and Trade: The AI Act amendment and new guidelines clarify the rules for using artificial intelligence, press release 29 July 2026 (in Czech)



