The EU Artificial Intelligence Act (Regulation (EU) 2024/1689) is the first comprehensive legal framework for regulating artificial intelligence (AI). Since becoming law on August 1, 2024, the AIA aims to ensure AI is safe, respects fundamental rights, and promotes trustworthy innovation across the EU. [digital-strategy.ec.europa.eu]
Core Approach
The AIA takes a risk-based approach and classifies AI systems into four risk levels (like a pyramid):
- Unacceptable risk — Banned outright
- High risk — Strictly regulated with obligations
- Limited risk — Transparency requirements
- Minimal/no risk — Largely unregulated (vast majority of AI applications)
[digital-strategy.ec.europa.eu]
Prohibited Practices (Unacceptable Risk)
Eight specific practices are banned because they threaten safety, livelihoods, or rights.
- Harmful manipulation or deception (e.g., subliminal techniques distorting behavior).Exploitation of vulnerabilities (e.g., targeting children or disabled people)
- Social scoring (leading to unfair treatment)
- Individual criminal risk prediction based solely on profiling (with exceptions)
- Untargeted scraping of internet/CCTV for facial recognition databases
- Emotion recognition in workplaces or education (with medical/safety exceptions)
- Biometric categorization inferring sensitive traits (race, politics, etc.)
- Real-time remote biometric identification in public spaces by law enforcement (narrow exceptions).
[https://ec.europa.eu/newsroom/dae/redirection/document/112367]
