February 2, 2025: The Day Prohibited AI Practices Became Illegal
The first major enforcement milestone of the EU AI Act has arrived. Here's what's now banned, who's affected, and how regulators will enforce these new rules.
A Historic Day for AI Regulation
February 2, 2025 marked a watershed moment in AI governance. For the first time in history, specific artificial intelligence applications are now categorically illegal across the European Union. This isn't a future deadline—it's happening now.
What's Now Prohibited?
Article 5 of the AI Act defines "unacceptable risk" AI systems that are completely banned:
- Social Scoring Systems: Government systems that evaluate citizens based on behavior or personality characteristics, leading to detrimental treatment
- Subliminal Manipulation: AI techniques that deploy subliminal components to materially distort behavior in ways that cause harm
- Exploitation of Vulnerabilities: Systems that exploit age, disability, or social/economic situation to distort behavior
- Predictive Policing (Individual): AI making risk assessments of individuals to predict criminal offenses based solely on profiling
- Untargeted Facial Recognition Scraping: Creating facial recognition databases through untargeted scraping from CCTV or internet
- Emotion Recognition in Workplace/Education: Inferring emotions in workplaces and schools (with narrow exceptions)
- Biometric Categorization: Categorizing individuals based on biometric data to infer sensitive attributes like race, religion, or sexual orientation
- Real-time Biometric ID in Public Spaces: By law enforcement, except for specific serious crime investigations with judicial authorization
Who Needs to Act Immediately?
Any organization currently using AI systems that fall into these categories must cease operations immediately. This includes HR departments using emotion AI in interviews, educational institutions monitoring student engagement via facial analysis, security firms using untargeted biometric databases, and marketing companies using subliminal AI techniques.
Penalties for Non-Compliance
Violations of Article 5 prohibited practices carry the highest penalties under the AI Act: up to 35 million euros or 7% of global annual turnover, whichever is higher. For context, this exceeds GDPR's maximum fines.
Guidelines Published
On February 4, 2025, the European Commission published official Guidelines on Prohibited AI Practices, providing legal explanations and practical examples to help organizations understand their obligations.
Next Steps for Enterprises
- Complete an immediate AI system inventory
- Review all systems against Article 5 prohibitions
- Decommission any prohibited systems
- Document the assessment process thoroughly
- Train staff on the new legal landscape
The era of unregulated AI in Europe is officially over.