AI Regulation & Governance

The regulatory landscape — EU AI Act, data privacy, model cards, and building compliant AI systems

AI Regulation & Governance

EU AI Act (2025-2026)

Risk-based framework:

Risk Level	Examples	Requirements
Unacceptable	Social scoring	Banned
High	Hiring, credit, medical	Assessment, human oversight
Limited	Chatbots	Transparency
Minimal	Spam filters	None

Penalties: Up to 35M EUR or 7% of global revenue.

Model Cards

Document describing an AI model:

Model: GPT-4o
Purpose: General text generation
Training: Internet text, cutoff Oct 2023
Performance: MMLU 88.7%, HumanEval 87.2%
Limitations: May hallucinate, has biases
Mitigations: RAG for factuality, human oversight

Governance Checklist

[ ] Named person responsible for AI governance

[ ] Model cards and data cards maintained

[ ] Regular bias audits

[ ] Incident response plan

[ ] Users informed when interacting with AI

[ ] Appeal process for AI decisions

Your Turn!

Classify under EU AI Act: job screening, pizza chatbot, social scoring, medical X-ray analysis, email spam filter.

✏️ Code Editor

Loading Python...

📤 Output

Write your solution and click "Run Code" to test it!

← Advanced Prompting Techniques Next: Evaluation & Benchmarks →