AI Act
EU Artificial Intelligence Act
In Swedish: "AI-förordningen"
Embedded Files
Written by Martin Ågren - May 26, 2026
FACTS about the LAW
The law is called the "AI Act".
Formal title in the document describing the law: "EU Artificial Intelligence Act".
Formal name of the law, legal designation: "Regulation (EU) 2024/1689 on harmonized rules for artificial intelligence"
The world’s first AI legislation.
The European Parliament passed the law in March 2024.
The law began to apply partially in 2024 with a gradual rollout in the following years.
The AI Act uses a risk pyramid to categorize and regulate AI. The higher the risk an AI system entails, the stricter the rules apply.
RISK PYRAMID
Risk pyramid for the AI act.
Image by Martin Ågren (generated via ChatGPT).
"Unacceptable Risk"
Prohibited to use AI in this way.
Example: Real-time facial recognition without specific justification.
"High Risk"
Strictly regulated. Authorization required.
Example: Recruitment
Example: Exams in school graded by AI
Example: Police use of AI, e.g. surveillance cameras
"Limited Risk"
Transparency required
Example: Chatbots for customer service
Example: AI-generated image
"Minimal Risk"
Free / unregulated use of AI in this category.
Example: Spam filters for email
For You as a MANAGER or LEADER
You should have a basic understanding of the AI Act and how the law affects your use of AI. Identify which risk level your AI systems may fall into and which requirements then apply.
For your AI policy: Clarify which AI tools may be used and how. Clarify who is responsible for compliance with the AI Act.
Page updated
Google Sites
Report abuse