How Technobel, a digital-focused training centre in Belgium, frames the use of artificial intelligence across its whole community, and why we chose to share our charter with others.
Where it started
In a digital training centre, generative AI is not a question for the future. Our learners already use it every day, and the companies that host them expect people who can work with these tools. Banning AI would have been unrealistic. Letting it in without any framework would have put at risk what gives training its value: the real acquisition of skills. So we chose to recognise AI as a learning tool, but a framed one.
A framework for the whole Technobel community
The charter is not only for learners. The same framework applies to trainers and to our learner support staff: the whole Technobel community shares the same principles and the same language when it comes to AI. This consistency matters. When the people who teach and support learners follow the rules they ask learners to follow, those rules become credible and part of our shared culture.
One core principle: AI supports learning, it does not replace it
The charter starts from a simple idea. AI can help people question, understand, rephrase, practise or explore a topic in more depth. It never replaces intellectual effort or personal practice. Learners remain fully responsible for the work they submit and must be able to demonstrate the targeted skills on their own.
Two complementary tools: usage status and transparency scale
The charter frames AI with two distinct tools, each with its own role.
Usage status: what is allowed. For each activity, the trainer states whether AI use is authorised, limited or prohibited. Limited use means AI may only be used for specific tasks set by the trainer, such as rephrasing instructions, generating ideas, checking language or structuring thoughts, without ever producing the expected work. AI is always prohibited during assessments, whatever their format. This lets us adapt AI use to the stage of the pathway, the learning objectives and the level of autonomy expected.
Transparency scale: what was actually done. To declare how AI was used, we rely on the four-level scale developed by the Université du Québec à Chicoutimi (UQAC), with its matching pictograms:
- No use: the work was produced without any AI involvement.
- Limited assistance: AI was used only for minor suggestions or corrections, such as spelling, grammar or small visual improvements.
- Shared assistance: AI made a significant contribution, working in partnership with the author.
- Major assistance: the content was mostly produced by AI, with minimal human input.
The status sets the rule, and the scale makes practice visible. Each production states the tool used and the level of AI contribution, with citations following APA standards. Using AI without declaring it is treated as concealment. With these two tools, everyone speaks the same language, and AI use becomes something that can be seen, discussed and assessed.
Protecting content and data, including in companies
The charter applies wherever training takes place: on site, remotely, and during work placements in companies. It is forbidden to submit course materials, source code, internal projects, placement data or host companies' information to any AI tool. No personal data may be entered either. This is essential for the trust of our company partners. The charter is grounded in the GDPR and the EU Artificial Intelligence Act.
Critical thinking and autonomy
The charter asks everyone to avoid becoming dependent on these tools and to check every piece of generated content systematically, since AI can make mistakes or invent information. AI output reused without understanding or checking has no learning value and exposes its user to errors they remain responsible for.
A formal and evolving commitment
For learners, signing the charter is a condition of access to training. Breaches lead to graduated sanctions, from a warning up to permanent exclusion, depending on how serious they are. A designated referent answers questions, and any doubt about whether a use is allowed must be cleared before acting. Because tools and regulations evolve quickly, the charter is reviewed at least once a year.
How we built it
The charter was co-designed with trainers, learner support staff and learners themselves. Several meetings allowed us to start from real practice: what people were already doing with AI, what genuinely helped learning and what raised concerns. Involving learners from the start was essential, because a rule you helped define is easier to understand and to respect. The charter has been in force since September 2026 across all our training pathways.
What we learned
Trainers were asking for it. The charter gives them a shared framework to validate and guide AI use activity by activity, instead of each of them improvising alone. More surprisingly, learners were asking for it too. Far from seeing the charter as a constraint, they found clear reference points in it: knowing what is allowed, what is not, and how to declare their use honestly. Uncertainty was more uncomfortable for them than an explicit rule. Since the charter has only been in force for a few weeks, we will draw a first assessment at its annual review.
A commons, made to be reused
Our charter itself builds on a commons: the UQAC transparency pictograms. In turn, we share it under a Creative Commons BY-NC-SA licence. Any training provider can reuse and adapt it to their own context, as long as they credit the source, do not use it commercially and share their version under the same terms.
Feel free to get in touch to discuss your own approach: we would be glad to learn from your experience.
Please log in or sign up to comment.