AI Literacy: What Does Article 4 of the EU AI Act Require?

Many rules of the EU AI Act won't become applicable until August 2026, but there's one obligation that is already in effect: ensuring AI literacy. Article 4 of the regulation has been applicable since February 2, 2025, and it applies to every organization that develops or uses AI systems.

In this article, we break down what AI literacy means in the context of the EU AI Act, who it affects, and how you can implement it in your company.

What Does Article 4 Say?#

Importantly, the article does not specify a concrete training program or exam. Instead, it formulates a results-based obligation: you must ensure that your people understand what they're doing when they use AI.

Who Does the AI Literacy Obligation Affect?#

The regulation distinguishes two main actors:

Providers: Organizations that develop or market AI systems. The strictest requirements apply to them — developers, testers, and support teams all need to be AI-literate.

Deployers: Organizations that use AI systems in their operations. Most SMBs fall into this category. The obligation still applies: employees must understand the AI tools they use.

What Does "AI Literacy" Mean in Practice?#

The regulation defines AI literacy as "the skills, knowledge, and understanding that allow providers, deployers, and affected persons to make an informed deployment of AI systems, and to be aware of the opportunities and limitations of AI, as well as the possible risks."

Translating this to practice, employees need to be literate in the following areas:

Basic AI Knowledge#

• What AI can and cannot do (understanding possibilities and limitations)
• The difference between AI-generated and human-created content
• Why you should never blindly trust AI output (hallucinations, biases)

Tool-Specific Knowledge#

• Basic understanding of how the specific AI tool works (at a functional, not technical level)
• What data the tool processes and how it handles it
• The tool's limitations and known error patterns

Responsible Use#

• The company's internal AI usage policies
• What data can and cannot be entered into AI systems (personal data, trade secrets)
• When human review is required before acting on AI output

Regulatory Awareness#

• Basic knowledge of the EU AI Act (its existence, main principles)
• Knowledge of obligations relevant to one's own role
• Knowing who to contact with questions or problems

How to Implement It in Your Company#

1. Assess the Starting Level#

Before organizing any training, it's worth assessing where your employees stand. A simple questionnaire can help:

• What AI tools do you use in your work?
• How did you learn to use them?
• Do you know what data the tool processes?
• Are you familiar with the company's AI usage policies?

2. Organize a Foundation Workshop#

The core of an AI literacy program is a well-structured internal workshop. It doesn't need to be an all-day affair. A 2-3 hour interactive workshop with the following structure makes a solid foundation:

First block (45 minutes): What is AI and how does it work?

• AI basics (non-technical approach)
• Generative AI: what it's good for and what it's not
• Hallucinations, biases, and limitations demonstrated with practical examples

Second block (45 minutes): Our tools in practice

• Overview of AI tools used at the company
• Examples of correct and incorrect usage
• Group exercise: solving a task with AI

Third block (30 minutes): Rules and responsibility

• The company's AI usage guidelines
• EU AI Act basics (what you need to know)
• Who's responsible, who to contact with questions

3. Create an Internal AI Usage Guide#

Alongside the workshop, it's worth preparing a short written guide that can be referenced at any time. It should include:

• The list of approved AI tools
• What tasks they may be used for
• What data must never be entered into AI systems
• Rules for reviewing AI output
• A contact person for questions or issues

4. Document the Training#

This step is less glamorous but important for compliance. Record:

• Who attended the training
• When the training took place
• What topics were covered
• Who conducted it

5. Embed It in Daily Operations#

AI literacy isn't a one-time event. The following should be built into how the company operates:

• Onboarding: every new employee receives AI training upon joining
• Biannual refresher: a short (1-hour) update on changes and new tools
• Access to the guide: make it easily accessible (e.g., internal wiki, shared drive)
• Ask-freely culture: employees should feel comfortable asking questions about AI usage

What Are the Penalties for Non-Compliance?#

Violation of the AI literacy obligation falls under the EU AI Act's general sanctions framework. While sanctions for deployers of minimal-risk systems are milder, the regulation allows for fines of up to EUR 7.5 million or 1.5% of annual turnover.

In practice, authorities will likely issue warnings first, especially for SMBs. But you shouldn't count on that: enforcement practices will crystallize over the coming years.

Summary#

AI literacy is not a bureaucratic burden — it's an opportunity. Companies where employees understand and consciously use AI are more efficient, make fewer mistakes, and better leverage the technology's potential.

The most important action items:

• Assess where your employees stand
• Organize a 2-3 hour foundation workshop
• Create an internal AI usage guide
• Document the training sessions
• Refresh knowledge regularly

The AI literacy obligation is already in effect. You don't need to be perfect, but you do need to start.