AI Act · Art. 4·In force since 2 February 2025

AI training is no longer a choice. It's a legal obligation.

Article 4 of the AI Act requires every company that uses artificial intelligence to ensure an adequate level of competence among its staff. We help you become compliant with a hands-on path, not a box-ticking course.

2 Feb 2025
The AI literacy obligation is already in force. It's not a future requirement.
2 Aug 2026
From this date national authorities begin enforcement and penalties.
Anyone using AI
It doesn't only concern AI developers: simply using ChatGPT, Claude or Copilot at work is enough.

In short

Since 2 February 2025, Article 4 of the AI Act requires every company that uses artificial intelligence to ensure an adequate level of staff training. The obligation applies to anyone using tools like ChatGPT, Claude or Copilot at work, not just to those who build AI. From 2 August 2026 national authorities can impose fines up to 35 million euros or 7% of global turnover. No certificate is required: what matters is real training, proportionate to each role and documented. Maverick AI helps companies become compliant with a practical 6-step method.

What Article 4 actually says

The AI Act (Regulation (EU) 2024/1689) is the world's first comprehensive law on artificial intelligence. Article 4 is short and direct:

Providers and deployers of AI systems shall take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf.

Article 4, Regulation (EU) 2024/1689 (AI Act)

In plain terms: if your company uses AI, you must make sure the people who use it know what they're doing. What it is, how it works, what risks it involves, what they can and cannot do with data. You don't need a certificate. You need real, documented training.

Are you obliged?

The obligation is broad. It covers three categories, and the third one concerns almost everyone.

Those who develop AI (providers)

If you build or customise AI systems, even just by integrating a model into your own product, you're a provider and the obligation applies to you.

Those who use AI in business (deployers)

If your organisation uses AI systems in its own processes under its own responsibility, you're a deployer. It's the broadest category.

Anyone using tools like Claude or ChatGPT

Marketing generating copy, HR doing screening, finance analysing data. If your team uses AI at work, the duty to train them is already yours.

What you need to do, concretely

The rule doesn't impose a specific course. It requires measures proportionate to role and risk. These are the four pillars.

1

Map where you use AI

Which tools run in your company, in which processes, with which data. Without this picture you don't know what to train or whom.

2

Train according to role

An executive, an analyst and a developer have different needs. Training must be calibrated to each person's skills, context and risk level.

3

Document what you do

You don't need a certificate, but you need proof. Just like with GDPR: training in itself isn't penalised, but its absence is the first thing an authority will challenge.

4

Keep it up to date

AI changes every month. One-off training isn't enough: you need an ongoing approach that keeps pace with new tools and risks.

How we make you compliant

We don't sell a recorded course to show in case of an inspection. We build a path around your real processes, one that genuinely trains people and leaves you with the documentation you need. It's the 6-phase method we use with all our clients.

01Assessment: we map processes, tools and skills
02Workshop: hands-on training on the team's real tasks
03Prompt & Skills Library: a library built on your own use cases
04Follow-up: a recall session to consolidate
05Adoption: support for everyday use
06Measurement and documentation: the proof of training delivered

Enforcement starts in August 2026

You have a window to become compliant before authorities start to monitor. It's worth using it.

Let's talk now

Frequently asked questions

Answers to the most common questions about the AI Act training obligation.

Yes. Article 4 of the AI Act has been in force since 2 February 2025 and requires providers and deployers of AI systems to ensure a sufficient level of AI literacy among staff. It's not a recommendation: it's a legal obligation that applies to anyone using AI in a professional setting.
Yes. The obligation doesn't only concern those who develop AI, but also those who use it (the deployers). If your teams use tools like ChatGPT, Claude or Copilot to work, your company must ensure they know how to use them consciously and safely.
From 2 August 2026 national supervisory authorities can monitor and impose penalties. The AI Act's penalties are proportionate to the severity and the size of the company, up to a maximum of 35 million euros or 7% of global turnover. As with GDPR, the absence of documented training is evidence of negligence.
No. The law doesn't require formal certifications or a dedicated governance structure. It asks for measures proportionate to role and risk, and recommends documenting the initiatives carried out internally. What matters is that the training is real, relevant and traceable.
There's no minimum duration. The rule asks that it be sufficient relative to the context. In practice that means calibrating it to different roles: basic awareness for those who use AI occasionally, more in-depth paths for those who use it intensively or build solutions.
We don't run theoretical box-ticking lessons. We work on your team's real processes and documents, with a 6-phase method that produces concrete skills, reusable assets and the documentation the AI Act requires. We specialise in Anthropic's Claude ecosystem and we build Maverick with the very tools we teach.

Become compliant with the AI Act

A hands-on path that genuinely trains your team and leaves you with the documentation you need. We start with a no-obligation call.

Mandatory AI training: the AI Act obligation (Art. 4)