EU AI Act
AI Act as European Union regulation
What is it?
The EU AI Act is a European Union regulation that establishes the world's first comprehensive law regulating artificial intelligence. As an EU regulation, it will apply directly in all member states without the need for implementation into national law.
What is this regulation about?
The main objectives of the AI Act are to:
- ensure the safety of people and the protection of fundamental rights,
- increase trust in AI,
- without stifling innovation.
The regulation governs how AI systems can be designed, deployed, and used within the EU.
Risk levels
The AI Act divides AI systems into categories depending on the level of risk they pose:
-
Unacceptable risk 🚫 (total ban on use):
- Systems that violate human rights, e.g.:
- real-time mass biometric surveillance,
- systems that manipulate human behavior (e.g., exploiting the vulnerability of children), so-called social scoring of citizens.
- Systems that violate human rights, e.g.:
-
High risk 🔴 :
-
Systems used in sensitive areas, e.g.:
- recruitment and HR,
- education (exams, grades),
- banking and creditworthiness,
- healthcare,
- justice.
-
👉 Permitted, but under very strict requirements, including:
- risk assessment,
- high-quality data,
- documentation,
- human oversight,
- cybersecurity.
-
-
Limited risk 🟡 :
-
Communication systems and content generation:
- chatbots,
- content generation systems (text, image, deepfake).
-
👉 Transparency obligation — the user must know that they are dealing with AI.
-
-
Minimal risk 🟢 :
-
Supporting systems:
- AI in games,
- spam filters,
- music or movie recommendations.
-
👉 No new obligations — complete freedom.
-
Who does the AI Act apply to?
- Companies from the EU and outside the EU if their AI is used in the Union,
- Manufacturers, suppliers, and users of AI systems, including big tech companies, startups, and public institutions.
Penalties
Violations are subject to very high financial penalties — up to tens of millions of euros or a percentage of global turnover.
EU AI Act vs. defense applications
If an AI system is used for military purposes, it is not subject to the AI Act.
(24) "If, and insofar as, AI systems are placed on the market, put into service, or used with or without modification of such systems for military, defence or national security purposes, those should be excluded from the scope of this Regulation regardless of which type of entity is carrying out those activities, such as whether it is a public or private entity."