Since February 2025, the rule has been in place: companies must ensure their employees working with AI systems have adequate competency. But now it gets serious. Starting August 2, 2026, national market surveillance authorities can check and enforce whether this requirement is met. For many companies, it's a wake-up call – because so far, training hasn't kept pace with AI deployment in many businesses.
Key Facts
- Deadline: August 2, 2026 – National authorities begin monitoring employee AI competency from this date
- No prescribed format or certification – The EU AI Act mandates no specific training or certificates, but requires tailored qualification
- External partners included – Service providers and freelancers working on behalf of the company must also be trained
- Compliance obligation since February 2025 – The requirement exists already; enforcement begins in roughly one year
What Companies Should Do Now
The first task: get an overview. Which AI systems are being used where? For what purposes? With what data? What risks emerge? Many companies should develop an AI Policy in parallel – an internal rulebook defining which applications are permitted, what data can be entered, how results are verified, and who is responsible.
"Companies should not view training as a one-time compliance exercise, but systematically and application-focused build AI competency. This is not only legally relevant but also helps unlock the actual value of AI."
This is how Joachim Pawlik, CEO PINKTUM, frames the challenge.
Tailored Training, Not One-Size-Fits-All
Training must match employees' concrete tasks. Someone using AI for draft text needs different skills than an executive integrating AI-driven analysis into decisions. A foundational course creates shared understanding – how AI systems work, where their limits lie, that even convincingly worded outputs can be wrong or biased. But after that, role- and application-specific training must follow.
Equally important: everyone must know which systems they can use for which tasks, what data can be entered, who reviews and approves results. In practice, this means clear rules – for example, which customer data or internal documents can go into an AI system and which cannot.
What This Means for Companies
The enforcement deadline is approaching: in about a year, authorities will check whether companies have done their homework. Anyone who hasn't yet conducted an inventory of their AI usage should do so quickly. At the same time, this is an opportunity: those who understand how AI works and where its limits lie avoid costly mistakes and deploy the technology more productively. Investment in genuine, tailored competency pays off twice – as compliance and as business value.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
