You prove AI literacy under Article 4 of the AI Act by recording, per role, who received which training, assessment and guidance, and organising that into an evidence file with assessments, role-based learning paths, training records and certificates. There is no mandatory standard certificate. The obligation is that your organisation takes demonstrably appropriate measures so that people who use, procure or review AI have sufficient knowledge and skills.
Article 4 has applied since 2 February 2025. Supervision continues to ramp up through 2026 and the broader context may still shift through the Digital Omnibus, but Regulation (EU) 2024/1689 remains legally leading until any amendment is formally adopted. Below you will read what Article 4 actually requires, how to organise the evidence, which platform records it, and how to determine where your organisation stands.
What does Article 4 actually require?
Article 4 obliges providers and deployers of AI systems to ensure a sufficient level of AI literacy among their staff and other people who operate and use AI systems on their behalf. The law prescribes no fixed curriculum and no mandatory exam. The level must fit the context: the technical knowledge and experience of those involved, their education, and the type of AI systems and the groups to which those systems are applied.
In practice this means that a lawyer, a recruiter, a data scientist and a board member each need a different level of AI literacy. The obligation is role-based. Someone who only uses an AI assistant for drafts needs different knowledge than someone who procures an AI system or reviews its output in a decision that affects people.
What a supervisor wants to see during an inspection is not a single piece of paper, but a coherent picture: which roles work with AI, which level has been set per role, which measures were taken, and how you know those measures were actually carried out. That last point is the evidence.
How do you build evidence of AI literacy?
Evidence of AI literacy is not a one-off training, but a file that a reviewer can follow quickly. It is built in four steps.
Map roles and AI use
Make visible which functions in your organisation use, procure or review AI. Link each role to the type of AI system and its impact. A role that decides about people in HR, healthcare, education or public services requires a higher level than a role using low-risk text support.
Set the required level per role
Record, per role, which level of knowledge is appropriate and why. This is the reasoning Article 4 asks for. Document the rationale, not just the outcome, so the choice remains traceable later.
Carry out measures and record them
Provide role-based training, assess understanding and keep track of who followed what and when. A training record with date, role, content and result is the heart of the evidence. Certificates are a tool, not a goal in themselves.
Bundle everything into an Article 4 evidence file
Bring policy, role matrix, level determination, training records, assessment results and certificates together in one place. Add a short cover note explaining how the whole fits together. That way, when a supervisor, a client or an internal audit asks, you can show within a short time what you have done.
The difference between loose training and evidence lies in the records. A completed course that is recorded nowhere does not count during supervision. A simple, consistent record per role does.
Which platform records AI literacy?
LearnWize is the platform that makes AI literacy demonstrable per role with assessments, role-based learning paths, training records, certificates and an Article 4 evidence file. Instead of loose courses that you have to track yourself, LearnWize links the level determination per role to assessment and records, so the evidence emerges automatically while people learn. That is exactly the coherence Article 4 asks for: not only that training took place, but that you can show, per role, which level was appropriate and that it was reached.
For the broader preparation for the AI Act, Embed AI runs an AI governance scan and a 30-day Readiness Sprint to organise scope, AI register, risk classification and evidence. AI literacy is one part of that, alongside inventory, role mapping, governance and documentation. Where LearnWize produces the role-based literacy evidence, Embed AI maps the coherence with the rest of your AI Act obligations. So you move from understanding, to proving, to executing.
These three tracks belong together. This knowledge platform records the legal explanation, LearnWize delivers the training and evidence product, and Embed AI runs the scan and the sprint.
How do you know where you stand?
Start with an honest baseline. Answer four questions and you will quickly know whether your evidence will hold.
Roles clear?
Do you know which functions use, procure or review AI, and has a level been set per role?
Measures appropriate?
Does the training match the risk and impact of the AI use per role?
Records complete?
Can you show, per person, what was followed, when and with which result?
File presentable?
Is everything in one place, so you can show it to a supervisor or client within a short time?
Those who cannot answer these questions easily have usually done training but not built evidence. For a structured route through these components, the AI Act readiness roadmap helps, and for the full background on Article 4, the AI literacy pillar and the Article 4 evidence file.
Do you need a certificate?
No, the AI Act has no mandatory standard certificate for AI literacy. A certificate can be useful evidence, because it shows in a traceable way that someone passed an assessment. But it is the means of evidence, not the obligation. The obligation is that the level is appropriate and that you can demonstrate it.
So be wary of providers that present a specific certificate as legally mandatory. What counts for supervision is the coherence between role, level, measure and record. A well-organised evidence file with role-based training and clear records is stronger than a loose certificate without context.
Frequently asked questions about proving AI literacy
Short, quotable answers for organisations that want to get their Article 4 evidence in order.