66 Exclusions The AI Act does not cover: 1. National security and defense. AI systems exclusively used for military or national security purposes are excluded. 2. Judicial and law enforcement cooperation with third countries. Thirdcountry public authorities or international organizations using AI for law enforcement or judicial cooperation, as long as they protect individual rights, are excluded. 3. Research and development (R&D). Activities related to R&D on AI systems or models before market placement or before placement into service are excluded. 4. Personal nonprofessional activity. AI systems deployed in purely personal, nonprofessional contexts are excluded (e.g., home security systems). 5. AI systems released under free and open-source licenses. Opensource AI systems (unless displaying unacceptable risk, high risk or being subject to limited risk rules) are excluded. However, “dual-use” systems (e.g., systems designed for both civilian and military use) and testing in real-world conditions (i.e., temporary testing of an AI system in real-world conditions to assess and verify its conformity with the AI Act) are within the scope. Enforcement and Penalties Compliance with the AI Act will be overseen by 27 national market surveillance authorities designated by member states and the European data protection supervisor. Member states must designate their national market surveillance authority by August 2, 2025. Key enforcement measures include: 1. Fines (not applicable before August 2, 2025) — for providers and deployers, violations can result in penalties up to €35 million or 7% of annual worldwide turnover, whichever is higher. 2. Withdrawal of the AI system from the market. 3. Restriction of the AI system’s availability. 4. Corrective actions to ensure the AI system is compliant.
RkJQdWJsaXNoZXIy MjgzNzA=