These prohibitions also cover practices like predictive policing and unauthorized facial image scraping, emphasizing the Act's broad scope.

Importantly, the Act applies to non-EU businesses operating within the EU, reflecting the 'Brussels effect' on global compliance.

The guidance outlines eight specific AI practices that are entirely prohibited, including harmful manipulation, social scoring, and real-time biometric identification.

Enforcement mechanisms for the Act will become operational on August 2, 2025, prompting organizations to prepare for compliance in the interim.

The EU AI Act's prohibitions on certain AI practices officially took effect on February 2, 2025, with guidance published shortly after on February 4, 2025.

Chapters I and II of the Act focus on prohibited AI practices and the promotion of AI literacy, which are crucial for compliance.

SMEs, which contribute significantly to Europe's economy, face challenges in adopting advanced technologies due to rigid regulations and limited digital capabilities.

AI literacy requirements under Article 4 mandate organizations to ensure staff possess sufficient AI knowledge, although the definition of 'sufficient' remains unclear.

This legislation includes a ban on facial recognition systems and other high-risk AI technologies, alongside stringent rules for AI transparency and accountability.

The European Commission has emphasized a broad interpretation of prohibitions to prevent circumvention and ensure high protection levels.

However, concerns have been raised that the heavy compliance burdens may stifle innovation, particularly for small and medium enterprises (SMEs).

Business operators are required to assess the capabilities of their AI systems to identify potential prohibitions and implement necessary mitigating measures.