Responsible AI Platform
❤️❤️❤️
Lesson 1 / 12

The risk pyramid

🔺

Risk-based approach

The AI Act works with a pyramid of four risk levels. The higher the risk an AI system poses to society, the stricter the rules. The majority of all AI systems fall under minimal risk — the law targets specific applications with a major impact.

🚫

Prohibited AI

Social scoring, manipulation, real-time biometrics

Art. 5
⚠️

High risk

HR, credit, education, justice, medical

Annex III
👁️

Limited risk

Chatbots, deepfakes, emotion recognition

Art. 50

Minimal risk

Spam filters, games, search algorithms

Free

▲ More regulation | Less regulation ▼

1 / 5