EU AI Act Classification: General-Purpose vs High-Risk + Annex IV Technical Documentation
EU AI Act (in force 2024): classifies LLMs into 4 categories — prohibited, high-risk, limited risk, minimal. Which category your FT model falls in = compliance budget. If high-risk: Annex IV, CE marking, conformity assessment. Mandatory if selling to EU market from Turkey.
Şükrü Yusuf KAYA
28 min read
Advanced
1. EU AI Act 4 Kategori#
| Kategori | Örnek | Compliance |
|---|---|---|
| Yasaklı | Subliminal manipulation, social scoring | Hayır, yasaklı |
| High-risk | Medical AI, recruitment AI, credit scoring | Annex IV + CE + conformity |
| Limited risk | Chatbot, deepfake | Transparency: "AI olduğunu söyle" |
| Minimal | Spam filter, video oyun AI | Voluntary code of conduct |
Cookbook'un kuralı: Use-case'ini önceden sınıflandır:
- E-ticaret chatbot → limited risk (transparency only)
- TR hukuk QA → debatable, danışmana sor
- Tıbbi triage → high-risk (Annex IV zorunlu)
- Çağrı merkezi sentiment → limited risk
- Banking copilot → debatable (creditworthiness etkisi varsa high-risk)
✅ Teslim
- Use-case'ini AI Act'a göre sınıflandır. 2) High-risk'se hukuki danışman al. 3) Sonraki ders: 18.2 — KVKK + Machine Unlearning.
Yorumlar & Soru-Cevap
(0)Yorum yazmak için giriş yap.
Yorumlar yükleniyor...
Related Content
Part 0 — Engineering Foundations
Welcome to the Fine-Tuning Cookbook: System, Stage Taxonomy, and the Reproducibility Contract
Start LearningPart 0 — Engineering Foundations
Reproducibility Stack: Seeds, cuDNN Flags, and Deterministic CUDA — End the 'Works on My Machine' Problem
Start LearningPart 0 — Engineering Foundations
Environment Pinning: uv + pyproject.toml, CUDA Version Matrix, and Container Recipes
Start LearningConnected pillar topics