Who in our organisation needs AI literacy training under Article 4?
Everyone who operates or uses an AI system for you, from the person drafting emails with a chatbot to the administrator who configures a screening feature, plus contractors doing the same. Not everyone needs the same depth. The duty is calibrated to skills, experience, education and context of use, so plan it role by role. The article names two groups, your staff and other persons dealing with the operation and use of AI systems on your behalf, and the Commission reads the second group broadly.
Step one: find the users
Start with people rather than a list of tools. For each team, ask what they use that drafts, summarises, translates, ranks, scores, predicts or recommends. The answers often include tools nobody bought deliberately: the assistant inside the email client, the summariser in the CRM, the ranking feature in the applicant tracking system. Every person who uses one of those for work is inside Article 4.
Step two: sort them into three groups
Users. People who use an AI tool to do their own job. Drafting, summarising, translating, searching. This is most of the organisation, and the baseline they need is the same: what the tool does, where it fails, what must never go into it, and who to ask.
Configurers and reviewers. People who set up an AI feature, choose its settings, or act on its output about other people. The recruiter reading a ranked shortlist. The finance lead who switched on anomaly detection. The marketing manager who set the tone rules for the drafting assistant. They need the baseline plus depth on what their tool is doing and what their own check adds.
Overseers and owners. Managers responsible for the teams above, and anyone who will own the acceptable-use policy or the review cycle. They need to be able to spot over-reliance and to know when a use has moved into a higher-risk category, because Article 26 attaches specific human-oversight duties to deployers of high-risk systems.
Step three: add the people who are not on the payroll
The Commission's Q&A on AI literacy says the phrase "persons dealing with the operation and use of AI systems on behalf of providers/deployers" covers people who are not employees but are broadly under your organisational remit, and gives contractors, service providers and clients as examples. It adds that people working for a service provider or contractor need to have the appropriate AI skills. The agency that runs your social media with an AI drafting tool, and the contractor who administers your HR platform, both belong on the list.
Step four: decide the depth per group, and write it down
The rewritten Article 4 asks you to take account of technical knowledge, experience, education and training, and the context the systems are used in. That is the calibration. A baseline session for the users, a second session for the configurers and reviewers, and a briefing for the overseers is a defensible shape for most organisations. The record should say who was in which group and why, because that reasoning is the part a regulator would want to read.
A worked example
A sixty-person engineering consultancy: fifty-two users of a chatbot and a document summariser, five project leads who read AI-generated risk flags in the planning tool, two HR staff who see ranked applicants, and one IT administrator who set all of it up. That is three groups at three levels of depth, with one record, which maps onto the half-day and full-day tiers.
Sources
- Regulation (EU) 2024/1689, the AI Act, Article 4 and Article 26, as amended by Regulation (EU) 2026/1744.
- European Commission, AI literacy: questions and answers, the passages on persons acting on your behalf and on contractors.
This is general information about the EU AI Act, not legal advice. Enforcement is proportionate and case by case, and your sectoral regulator may issue guidance that applies to you.
