From 2 February 2025, the first rules of the EU Artificial Intelligence Act apply. One of them is short and easy to overlook: Article 4, on AI literacy. It does not only concern companies that build AI. Any organisation that uses AI systems in its work is in scope. This article explains what the text says as it stands on 2 February 2025, who it binds, and what a training team can sensibly do first.
This is not legal advice. It describes the law as it stood on 2 February 2025; see the update note at the end.
The rule in the original text
The AI Act is Regulation (EU) 2024/1689. It entered into force on 1 August 2024 and applies in stages. Chapters I and II, which include Article 4 and the prohibited practices in Article 5, apply from 2 February 2025. Most other rules follow from 2 August 2026.
Article 4, as adopted in 2024, reads:
Providers and deployers of AI systems shall take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf, taking into account their technical knowledge, experience, education and training and the context the AI systems are to be used in, and considering the persons or groups of persons on whom the AI systems are to be used.
The Act also defines AI literacy (Article 3(56)): the skills, knowledge and understanding that allow providers, deployers and affected persons to make an informed deployment of AI systems, and to gain awareness of the opportunities and risks of AI and the possible harm it can cause.
Who it binds
Article 4 binds two roles defined in the Act:
- Providers develop an AI system, or have one developed, and place it on the market or put it into service under their own name.
- Deployers use an AI system under their authority, except in a purely personal, non-professional activity.
In practice, an organisation whose staff use AI tools at work is likely to be a deployer. The duty covers its staff and "other persons" who operate or use AI systems on its behalf, such as contractors. It is a duty on organisations, not on individual employees, and the text does not prescribe a course, an exam or a certificate.
Three phrases in the text shape what is reasonable: "to their best extent", "a sufficient level" and the factors to take into account (what people already know, the context of use, and the people the AI is used on). A team that drafts letters with a general-purpose chatbot needs something different from a team that uses AI to screen applications.
Sensible first measures for a training team
The text leaves the choice of measures to the organisation. For a training or HR team, a proportionate start could look like this:
- Make an inventory. List the AI systems in use, who uses them and for what. Include tools that staff use on their own initiative.
- Sort by role and risk. Separate general users from people whose AI use affects others, for example in recruitment, assessment or customer contact.
- Set simple rules. Write down which tools may be used, which data may not go into them, and who to ask when in doubt.
- Train for the task, not the theory. Short sessions per role work better than one general lecture. Let people practise with realistic tasks: check an answer that sounds right, spot missing context, decide when a person must decide.
- Keep a record. Note who took part in what and when, and review it when tools or tasks change.
Practice matters because the risks of AI tools show up in use. A short scenario in which an assistant gives a confident but wrong answer teaches more than a slide that says "AI can make mistakes". This is where scenario-based learning fits: realistic situations, choices with consequences, feedback and a second try.
What Article 4 does not say
The original text does not set a deadline beyond its date of application, and it does not create a certificate or a register. It also does not turn every training provider into a regulated party: a training provider is in scope because it provides or uses AI itself. Separate, stricter duties apply later to high-risk AI systems, such as the human-oversight duties of deployers in Article 26.
Our approach to AI in learning is set out on our page on AI literacy and applied AI. If you want to talk about practice scenarios for your teams, contact us.
Update, August 2026
Article 4 was amended in 2026 by the Digital Omnibus on AI, Regulation (EU) 2026/1744, in force since 27 July 2026. The duty to "ensure … a sufficient level of AI literacy" was replaced by a duty to take measures to support the development of AI literacy, and the new text states that no specific level has to be guaranteed. Read what changed in AI literacy under the AI Act: Article 4 after the 2026 amendment.
Sources
- Regulation (EU) 2024/1689 (Artificial Intelligence Act) on EUR-Lex: Article 3(3), 3(4) and 3(56) (definitions), Article 4 (original text), Article 26, Article 113 (dates of application).
- Regulation (EU) 2026/1744 (Digital Omnibus on AI) on EUR-Lex, for the update note.
