EU AI Act · Article 4

AI literacy training
is now the law.

Since 2 February 2025, organisations that provide or use AI systems in the EU must ensure their staff have a sufficient level of AI literacy. This guide explains what Article 4 of the EU AI Act requires, who it applies to, and how to build a training program that satisfies it.

Book a Demo → See the program →
What the law says

Article 4 is short,
its reach is wide.

It requires providers and deployers of AI systems to ensure, to their best extent, a sufficient level of AI literacy in their staff and anyone operating AI on their behalf.

  • Covers both providers of AI and deployers, the organisations that simply use it
  • If your staff use ChatGPT or Copilot at work, you are a deployer and the duty applies
  • In force since 2 February 2025, one of the first duties to reach ordinary businesses
  • Reaches non-EU companies whose AI systems are used in the EU market
Art. 4

In plain terms

If your people use AI at work, your organisation is expected to train them to use it competently and responsibly, at a depth that matches their role.

What "sufficient" means

Literacy is role-based,
not one-size-fits-all.

The Act judges sufficiency in context: each person's role, the AI they work with and the people it affects. In practice that maps to four levels of literacy across the workforce.

Broad-workforce AI literacy The broad workforce

Everyday fluency

Practical use of everyday AI tools, an understanding of core concepts and limits, and grounding in responsible use.

Operator AI literacy Operators & power users

Applied skill

Deeper skill applying AI to real workflows, plus awareness of data protection and accuracy risks.

Technical AI literacy Technical specialists

Engineering depth

Engineering-level understanding of the AI systems they build, integrate and ship into production.

Leadership AI literacy Leaders & governance

Strategy & risk

Command of AI strategy, risk, ethics and the regulatory landscape, including the Act itself.

This is exactly the shape of a persona-based program, and why Kubicle structures AI training for organisations around four workforce personas rather than one generic course.

The build

How to build a compliant
AI literacy program.

A defensible program is a process, not a one-off course. These five steps turn the Article 4 obligation into evidence you can show a regulator.

1
Map

Who uses AI, and how

Inventory the AI systems in use and the roles that touch them, defining who the obligation covers.

2
Benchmark

Current AI literacy

Assess the workforce against capability levels. A documented baseline is your starting evidence.

3
Assign

Role-calibrated paths

Match each role to the right depth: fundamentals for everyone, technical and governance where needed.

4
Anchor

To your AI policy

Tie pathways to your own AI use policy and code of conduct.

5
Evidence

Record, report, refresh

Keep completion records, report coverage to compliance, and refresh as tools and regulation evolve.

The wider Act

Literacy is first.
More duties phase in.

AI literacy is the obligation most organisations meet first, but the Act rolls out further duties through 2027. Knowing the sequence helps you plan beyond training.

AI literacy becomes mandatory

Prohibited AI practices are banned across the EU, and the Article 4 duty to ensure a sufficient level of AI literacy takes effect for every provider and deployer.

February 2025

General-purpose AI rules begin

Providers of foundation and generative AI models take on transparency, documentation and governance obligations, with knock-on duties for the businesses deploying them.

August 2025

High-risk systems bite

The bulk of the high-risk AI system requirements start to apply: risk management, data governance, human oversight and ongoing monitoring.

2026

Final deadlines land

Extended deadlines close for high-risk AI embedded in regulated products that require third-party conformity assessment.

2027

For the full picture, read our EU AI Act summary and timeline and what the EU AI Act means for your business.

Go deeper

The EU AI Act,
from every angle.

Solution

AI training for organisations

The managed, persona-based program that operationalises Article 4: benchmarking, role-calibrated pathways and audit-ready reporting.
Program

AI literacy training program

The dedicated AI literacy curriculum: fundamentals, responsible use and applied fluency for the modern workforce.
Guide

EU AI Act summary, requirements and timeline

The whole Act in one read: risk tiers, key obligations and the enforcement timeline to 2027.
Guide

What the EU AI Act means for your business

A practical countdown to compliance for business leaders, from literacy to high-risk systems.
Guide

Creating an AI use policy

A comprehensive enterprise guide to writing the AI policy your training program should teach.
Guide

AI codes of conduct in use policies

How codes of conduct translate policy into the day-to-day behaviour Article 4 literacy supports.
Frequently asked

EU AI Act training,
answered.

Still have a question about compliance? Book a call and a program designer will walk your compliance and L&D teams through it.

Book a Demo →
Does the EU AI Act require AI training? +

Yes. Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures to ensure, to their best extent, a sufficient level of AI literacy in their staff and other people operating or using AI systems on their behalf. The obligation has applied since 2 February 2025.

Who does the Article 4 AI literacy obligation apply to? +

It applies to organisations that provide AI systems and to organisations that deploy or use them in the EU, regardless of company size or sector. If your staff use AI systems in the course of their work, including generative AI tools, the literacy obligation is relevant to you. It also reaches non-EU companies whose AI systems are used in the EU market.

What counts as sufficient AI literacy under the EU AI Act? +

The Act ties sufficiency to context: staff need skills, knowledge and understanding appropriate to their role, the AI systems they work with, and the people affected by those systems. In practice that points to role-based training: broad AI fundamentals and responsible use for the general workforce, and deeper technical, risk and governance training for people building, buying or overseeing AI.

How do organisations evidence AI literacy compliance? +

Regulators expect organisations to be able to show what training was provided, to whom, and how it maps to roles and AI use. A defensible program includes a workforce skills assessment, role-calibrated training paths, completion and assessment records, and periodic refreshes as tools and regulation evolve.

Is there a penalty for ignoring the AI literacy requirement? +

Article 4 has no standalone fine attached, but the literacy obligation is taken into account in enforcement of the wider Act, where penalties are substantial, and national market surveillance authorities can act on non-compliance. Just as importantly, courts and regulators can treat inadequate AI literacy as an aggravating factor when AI use causes harm.

Get compliant

Stand up an Article 4-ready
AI literacy program.

Kubicle benchmarks your workforce, assigns role-calibrated AI literacy pathways and gives your compliance team the records to evidence it. Scope it in a 20-minute call.

Book a Demo → See the program →