How does your
AI literacy stack up?
Art. 4 EU AI Act requires a sufficient level of AI literacy among your staff. This check shows honestly where you stand – measured against what the EU Commission actually expects. No sign-up, no data transfer.
An honest read in four steps
We ask deliberately specific questions – because a sound first self-assessment only works with precise questions. The result is an orientation, not legal advice.
Does your company use or develop AI?
How was AI literacy built up in the team?
This refers to the people who actually work with AI.
What does this competence cover – or is it just about operating the tool?
Multiple selection. These are the content points the EU Commission names for "sufficient".
Do you operate or develop a high-risk AI system?
e.g. AI in HR/recruiting, credit scoring, critical infrastructure. Unsure? The AI Act Check clarifies it precisely.
As of today Art. 4 does not apply to you – you currently use no AI. As soon as an AI tool is used at work (including ChatGPT & co.), the AI literacy duty kicks in. A good opportunity to be prepared.
You are well positioned: systematic, documented competence that covers the role-specific dimension, the risks, and the legal framework. That matches what the EU describes as "sufficient" and is defensible to a regulator – though Art. 4 has no formal proof of conformity.
ℹ AI changes fast – keep the competence current and document refreshers.
Art. 4 applies to you – and measured against the EU Commission's requirements there is a gap. Missing: –.
✓ This is exactly where structured training comes in: role-specific, covering risks and the legal framework, demonstrably documented – not just "how do I operate the tool".
⚠ Stricter: for high-risk AI, trained, competent human oversight is mandatory (Art. 26 / 14) – the Omnibus did not soften this. Here training is a duty, not just a recommendation.
ℹ Art. 4 applies since February 2025 (enforcement from August 2026). A softening to a mere support duty is planned via the Omnibus but not yet adopted – the content bar (risks, law, role-fit) remains the benchmark.
How is "sufficient" measured?
The EU Commission deliberately sets no rigid requirement (no number of hours, no mandatory certificate), but names clear content elements of good AI literacy: general understanding (what AI is, opportunities and risks), clarifying the role (provider/deployer), risks & mitigations of the systems in use, tailoring to prior knowledge and role, and law & ethics (AI Act).
What counts as "defensible" to a regulator is a tiered, role-specific approach – foundations for all plus depth per task. The AI Office maintains a continuously updated "Living Repository of AI Literacy Practices".
Status June 2026. The AI Act is being amended via the "Digital Omnibus" (political agreement May 2026, not yet in force). We keep the check up to date.
Good to know: AI training is often eligible for funding – incl. a wage subsidy. Our Funding Check shows you. Legal basis: Art. 4 EU AI Act & the EU Commission's AI literacy Q&A.
What "sufficient AI literacy" means
The EU Commission prescribes no number of hours, but clear content. That's what this check measures against – and what we build our training around.
General understanding
What AI is, how it works, which systems are in use – opportunities and risks.
Risks & mitigations
The concrete risks of the systems in use and how to address them.
Role & tailoring
Provider or deployer? Foundations for all, depth depending on task and prior knowledge.
Law & ethics
AI Act, data protection and governance – not just operating the tools.