Key Takeaways
- Article 4 of the EU AI Act has required companies using AI systems to ensure staff AI literacy since 2 February 2025.
- The obligation covers everyday tools like ChatGPT, Copilot, or Gemini, not just custom-built or high-risk AI systems.
- No specific training format is mandated, and structured AI training in Germany can be funded through the Bildungsgutschein.
Table of Contents
What does Article 4 of the EU AI Act require?
Article 4 requires providers and deployers of AI systems to take measures ensuring their staff have a sufficient level of AI literacy. This isn’t about teaching employees to code. It’s about understanding how the AI systems they use actually work, including their capabilities, risks, and limitations.
The regulation defines AI literacy as the skills, knowledge, and understanding that allow staff to deploy AI systems in an informed way and to be aware of the opportunities, risks, and possible harms involved. That includes basic knowledge of how a system functions, relevant legal frameworks such as data protection, and the ability to recognize bias or flawed AI-generated outputs.
The requirement is broader than most companies assume. It applies to everyday generative tools such as ChatGPT, Microsoft Copilot, or Google Gemini as soon as they’re used in a professional context, not only to complex or purpose-built AI systems.
When did the AI literacy requirement take effect?
Article 4 is part of Regulation (EU) 2024/1689, which entered into force on 1 August 2024. Unlike many other obligations under the EU AI Act, including the extensive rules for high-risk AI systems, Article 4 became applicable much earlier, on 2 February 2025.
That makes the AI literacy requirement one of the first provisions of the regulation to take direct effect. Companies that haven’t yet put measures in place are not currently meeting the legal requirement, regardless of how advanced their AI usage is.
Who does Article 4 apply to?
The scope of Article 4 is intentionally broad and covers two groups that together include most companies operating in the EU.
Providers and deployers of AI systems
The requirement applies both to companies that develop AI systems or place them on the market under their own name, and to any company that uses AI systems as part of its professional activities. The second group covers the vast majority of businesses, regardless of industry or size.
People working on a company’s behalf are also included. That can extend to contractors, service providers, and in some cases even clients accessing a company’s AI-powered systems.
Everyday tools count too
A common misconception is that AI literacy obligations only apply to companies building their own AI models or running high-risk systems. In practice, widely used tools fall under Article 4 as well, including:
- Generative tools like ChatGPT, Claude, or Gemini used for writing and research
- AI-powered recruiting and applicant screening software in HR
- Automated customer communication through chatbots
- Internal data analysis tools with AI-driven features
Teams that assume their AI usage is too limited to be relevant often underestimate how far the requirement actually reaches.
What happens if a company doesn’t comply?
Article 4 doesn’t carry a standalone fine, unlike the rules on prohibited AI practices or high-risk systems. That doesn’t mean non-compliance is risk-free.
If untrained staff cause harm through flawed AI-driven decisions or data protection breaches, the company can face civil liability. Regulators are also likely to treat a lack of AI literacy measures as an aggravating factor in any wider investigation into other EU AI Act violations.
In practice, the literacy requirement is less a one-time compliance box to tick and more an ongoing process. New AI tools introduced into a company create new training needs, and a single onboarding session isn’t enough on its own.
How to build AI literacy in practice
The regulation doesn’t prescribe a specific training format, which gives companies flexibility but also leaves the implementation open. A layered approach tends to work well: baseline training for all staff, with deeper, role-specific content for anyone working directly with AI systems.
What a compliant training should cover
A solid AI literacy program typically addresses:
- Fundamentals: how the AI systems used in the company actually work
- Legal context: data protection, copyright, and accountability when using AI
- Risk awareness: recognizing bias, hallucinations, and flawed outputs
- Practical application: using AI tools safely and productively in day-to-day work
L&D teams often face the added challenge of documenting this training in a way that holds up if questioned later. Elena Giannetti, a graduate of the AI for Business Course, solved exactly this problem as an L&D manager: she built an automated training management system using Airtable, Slack, Make, and HeyGen to track staff training progress, a real example of how AI literacy efforts can be organized at scale inside a company.
Funded training options in Germany
Structured AI training doesn’t have to come entirely out of a company’s own budget. Individual employees, for example those retraining or moving into a more AI-focused role, may qualify for a Bildungsgutschein, which can cover course fees up to 100 percent.
WBS CODING SCHOOL’s AI for Business Course builds exactly this kind of practical literacy: over 8 weeks, it covers AI fundamentals, legal and ethical considerations, and hands-on use of generative AI tools, content directly aligned with what Article 4 asks companies to ensure.
For teams based in Germany that prefer German-language training, the KI Kompakt Course covers the same ground in a part-time, morning-only format with weekly intake, making it easier to fit around existing work schedules.
Frequently asked questions about EU AI Act Article 4
Do I need to formally test employees’ AI literacy?
No, Article 4 doesn’t require certification or formal testing of AI literacy. The obligation is to take appropriate measures to build competence, not to prove it through exams or certificates. That said, documenting completed training is a good idea, so the company can demonstrate compliance if it’s ever questioned.
Does the requirement apply to small companies too?
Yes, company size doesn’t affect Article 4. A five-person startup is subject to the same basic obligation as a company with thousands of employees. What matters is whether the company uses AI systems, not how big it is.
Is a one-off ChatGPT walkthrough enough to comply?
A short, one-time walkthrough rarely covers what Article 4 actually requires. The obligation is about understanding how a system works, its risks, and the legal context around it, not just how to operate the tool. Since AI tools used inside companies keep evolving, training also needs to be revisited regularly rather than delivered once and forgotten.
Related blogs
- AI skills: 5 competencies for your job security
- Best AI courses in Germany
- How to get a Bildungsgutschein
Conclusion
Article 4 of the EU AI Act is already in effect, and it applies far beyond companies building their own AI systems, reaching anyone using AI tools at work. Structured, documented training is the most practical way to meet the requirement while building a genuinely more capable team. WBS CODING SCHOOL’s AI for Business Course covers this exact ground in 8 weeks, or in German through the KI Kompakt Course, both fully fundable through a Bildungsgutschein.
