AI Act | Risk Classification

AI Act | Risk Classification

Note: This document is under development. Content will be added after finalizing the Czech version.

See Czech version: docs/cs/legislativa/ai-act/03-klasifikace.md

---

Summary

Detailed guide for classifying AI systems according to AI Act risk categories.

Risk Categories

Category	Examples	Obligations
🔴 Prohibited	Social scoring, subliminal manipulation	STOP - do not use
🟠 High-Risk	Credit scoring, employment, health	Full compliance
🟡 Limited Risk	Chatbots, deepfakes, emotion recognition	Transparency
🟢 Minimal Risk	Spam filters, analytics, recommendations	Minimal

Classification Decision Tree

Is AI in prohibited practices list (Art. 5)?
├─ YES → PROHIBITED
└─ NO ↓

Is AI a safety component or in Annex III areas?
├─ YES → HIGH-RISK
└─ NO ↓

Does AI interact directly with users?
├─ YES → LIMITED RISK (transparency)
└─ NO → MINIMAL RISK

---

Full content pending - Czech version finalization in progress