📰 Full Story

The European Commission has published the detailed implementation rules for the AI Act.

Risk-Based Classification

The Act categorizes AI systems into four risk levels:

  1. Unacceptable Risk: Social scoring, real-time remote biometric identification — fully banned
  2. High Risk: Medical diagnosis, autonomous driving, critical infrastructure — strict compliance required
  3. Limited Risk: Chatbots, deepfakes — users must be clearly informed
  4. Minimal Risk: Spam filters, recommendation systems — voluntary code of conduct

Global Impact

This is the world’s first comprehensive AI regulatory framework and is expected to produce a “Brussels effect” similar to GDPR, prompting other countries to follow suit with legislation.

The Act is expected to take effect in July 2026.

Source: European Commission, Financial Times