EU AI Act

The EU AI Act (Regulation (EU) 2024/1689) is the world's first comprehensive legal framework for artificial intelligence. It follows a risk-based approach, classifying AI systems into four categories: prohibited AI, high-risk AI, limited-risk AI, and minimal-risk AI.

• Prohibited AI: Certain applications such as government social scoring or manipulative AI are fully banned.
• High-risk AI: Systems in critical sectors (education, employment, law enforcement, biometric identification) are subject to strict requirements.
• Transparency: Users must be informed when they are interacting with AI (e.g., chatbots).
• General-purpose AI (GPAI): Special requirements for providers of large AI models such as GPT or Gemini.

The EU Commission published the first draft of the AI Act in April 2021. After intensive negotiations in the European Parliament and Council — particularly on real-time biometrics and GPAI models — a political agreement was reached in December 2023. The AI Act was adopted by the European Parliament on 13 March 2024 and published in the EU Official Journal on 12 July 2024 (OJ L 2024/1689), entering into force on 1 August 2024 (20 days after publication per Art. 113). Requirements apply gradually: prohibitions from February 2025, GPAI rules from August 2025, and high-risk requirements from August 2026 and 2027.

The AI Act applies to providers placing AI systems on the EU market or putting them into service, operators of AI systems established in the EU, and providers/operators outside the EU where the AI output is used within the EU. Exceptions exist for AI for military purposes and purely private use.

• Risk classification: Classification of all AI systems by risk level before market placement.
• High-risk requirements: Conformity assessment, technical documentation, human oversight, robustness, accuracy.
• GPAI obligations: Technical documentation, copyright compliance, transparency; for systemic risks: model evaluations, incident reporting.
• Prohibitions: Cognitive behavioral manipulation, social scoring, real-time biometrics in public spaces (with narrow exceptions).
• Market surveillance: Establishment of national authorities and an EU AI Office.
• Fines: Up to EUR 35 million or 7% of global annual turnover.

1 correction:
- AI Act: effective_date and key_dates contain incorrect dates