Advanced · Team and organisation · module 116 of 155
The AI Act in practice
Understand how the European AI Act classifies applications and what that means for a team that uses AI.
After this lesson
You explain the risk levels of the AI Act, recognise when an application needs extra attention and know where to look up the current rules.
In view · How it connects
The AI Act in three layers
A small number of practices, such as a social score that puts people at a disadvantage.
Strict requirements in areas such as recruitment, education and essential services.
People know they are talking to AI; organisations support AI literacy.
Part 1 of 3 · Prohibited
Your preference for less movement is on. Choose each part yourself.
A law that works with risk levels
The European AI Act (Regulation (EU) 2024/1689) classifies AI applications by risk. A small number of practices are prohibited, such as a social score that disadvantages people or manipulation that exploits vulnerabilities; those prohibitions have applied since February 2025. High-risk applications, for example in recruitment, education or access to essential services, are subject to strict requirements. For many other applications, it is mainly transparency obligations that apply, and for simple uses such as a spam filter there are hardly any specific rules.
What it means if you use AI
The law distinguishes between providers, who develop an AI system or place it on the market, and deployers, who use it under their own authority. So even if you use an existing tool, you have obligations. For example, providers and deployers must take measures to support the AI literacy of those who work with AI. Transparency obligations have applied since August 2026: people must know that they are talking to an AI system, unless that is obvious, and deepfakes must be recognisable as such.
How to approach it as a team
Make a list of the AI applications your team uses, with purpose, data and who deploys them. For each application, check whether it could fall into a prohibited or risky category, for example when AI has a say in decisions about job applicants. After an amendment to the law in 2026, most requirements for high-risk applications only take effect in December 2027, but it is wise to know now which applications are concerned. This lesson is not legal advice. Read the current text on EUR-Lex and seek advice if in doubt (lessons 28 and 93).
New practice example
From explanation to a usable result.
The AI Act classifies AI applications by risk. Even if you use an existing tool, you have obligations.
Fictional practice situationFictional: a non-profit (vzw) wants to use three AI applications: a chatbot for opening hours, a tool that helps write newsletters and a system that decides which applicants get priority for a social service.
View a possible solution
This is how it can be done
Chatbot: visitors must know that they are talking to an AI system. Newsletter tool: few specific rules, but internal agreements and training. Priority for social services: affects access to essential services and may count as high-risk; do not introduce it without legal advice, and a human decides.
Why this works
The same organisation can have applications with very different risks. An inventory per application shows where extra care is needed.
Make it slightly more difficult
In the current text on EUR-Lex, find the annex listing high-risk applications. Which domain is closest to your work?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
The same organisation can have applications with very different risks. An inventory per application shows where extra care is needed.
A step further
In the current text on EUR-Lex, find the annex listing high-risk applications. Which domain is closest to your work?
Try & discover · no obligation
Become the source checker yourself
Only mark statements that the source does not support. Then check your selection.
Fictional notes after a team meeting about the European AI Act. This is an exercise, not legal advice: check specific questions in the current official text.
Select which sentences you would correct above. The answer remains hidden until you check your selection.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Help me make an overview of the AI applications in a fictional team in a municipal department: a chatbot on the website for general questions, a tool that summarises meeting minutes and an idea to have AI rank job applications. For each application, make a row with purpose, data involved, who deploys it, which AI Act risk level may apply and which question we should ask a lawyer. Do not give a final legal judgement.
Apply it
A school wants to use an AI tool that automatically decides which pupils are admitted to a study track. The supplier says the tool is “AI Act-proof”.
Which approach best fits this situation?
- Use the tool, because the supplier bears all responsibility.
- Use the tool first as a trial for half of the pupils.
- First have the application legally assessed, because admission to education may fall under high-risk, and make sure people decide.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
Your team deploys a chatbot from a large provider on its own website. What is correct?
- Only the provider has obligations, you do not.
- You have obligations too, for example around AI literacy, and visitors must know that they are talking to an AI system.
- The AI Act only applies to companies outside Europe.
Extra practice
Make an inventory of the AI applications in your team or organisation. Mark which applications have a say in decisions about people, and note whom you can ask the legal questions.
Chatbot: general questions, no personal data needed; visitors must know that they are talking to an AI system. Minutes tool: internal notes; agreements on data and training for the users. Ranking job applications: may count as high-risk because it concerns recruitment; do not introduce without legal advice, and a person continues to make the decision. Question for the lawyer: which obligations apply to us as a deployer?
