AI Model Wiki

    Skip to main content
    12 min read
    Lesson 1 of 5

    The EU AI Act: World's First Comprehensive AI Law

    From the course AI Regulation & Policy Landscape

    The European Union's AI Act, which entered into force in August 2024, is the world's first comprehensive legal framework for artificial intelligence. It establishes a risk-based approach that will influence global AI regulation.

    Risk Categories

    The EU AI Act classifies AI systems into four risk levels:

    1. Unacceptable Risk (Banned) — social scoring by governments, real-time biometric surveillance in public spaces (with exceptions), manipulative AI targeting vulnerabilities, emotion recognition in workplaces and schools
    2. High Risk (Strict Requirements) — AI in critical infrastructure, education, employment, credit scoring, law enforcement, migration, and justice
    3. Limited Risk (Transparency) — chatbots, deepfakes, and AI-generated content require disclosure
    4. Minimal Risk (No Restrictions) — spam filters, video game AI, recommendation systems

    High-Risk AI Requirements

    High-risk systems must comply with extensive requirements:

    • Risk Management System — ongoing identification and mitigation of risks
    • Data Governance — training data must be relevant, representative, and error-free
    • Technical Documentation — detailed records of system design, development, and testing
    • Record-Keeping — automatic logging of system operations for traceability
    • Transparency — users must understand AI system capabilities and limitations
    • Human Oversight — meaningful human control over AI decisions
    • Accuracy & Robustness — systems must perform consistently and resist manipulation

    General-Purpose AI Models (GPAI)

    Special provisions for foundation models like GPT-4 and Gemini:

    • All GPAI — technical documentation, training data transparency, copyright compliance
    • Systemic Risk GPAI (10^25 FLOPs threshold) — additional model evaluation, adversarial testing, incident reporting, and energy consumption tracking

    Timeline & Enforcement

    • February 2025: Banned practices provisions apply
    • August 2025: GPAI and governance provisions apply
    • August 2026: Full high-risk system requirements apply
    • Fines up to €35 million or 7% of global annual turnover

    Key Takeaways

    • •The EU AI Act uses a four-tier risk classification system
    • •High-risk AI requires risk management, data governance, and human oversight
    • •General-purpose AI models have specific transparency and evaluation requirements
    • •Fines reach up to €35 million or 7% of global annual turnover
    • •Full enforcement of high-risk requirements begins August 2026
    0 of 5 lessons marked complete

    Progress is kept in this browser only.