Zum Inhalt springen
>_<
AI EngineeringWiki

Prohibited AI Practices

Compliance · 4 min · Updated: Feb 2026

Attention: In Force since 2 February 2025

Article 5 of the EU AI Act prohibits certain AI practices absolutely. No balancing, no exceptions — absolute prohibitions with penalties up to EUR 35M.

The 6 Prohibited Practices

1. Subversive Manipulation (Art. 5(1)(a))

What is prohibited: AI systems that lead people to harmful decisions through "subliminal techniques" or deliberate deception.

Example: An AI tool that manipulates impulsive purchasing behavior through hidden triggers.

2. Social Scoring (Art. 5(1)(b))

What is prohibited: AI systems that evaluate persons based on their social behavior or non-legally justified criteria.

Example: A system that assesses creditworthiness based on social media activity.

3. Biometric Categorization (Art. 5(1)(c))

What is prohibited: Use of biometric data to categorize persons in real-time at public places.

Example: Real-time facial recognition in public places to detect "suspicious behavior".

4. Emotion Recognition in Workplace (Art. 5(1)(d))

What is prohibited: AI systems for emotion recognition in the workplace and educational institutions.

Example: Webcam analysis during job interviews to assess "trustworthiness" or "nervousness".

5. Untargeted Facial Data Collection (Art. 5(1)(e))

What is prohibited: Creating or expanding databases through untargeted collection of facial images from the internet.

Example: Scraping social media profiles to create face recognition databases.

6. Government Social Scoring Systems (Art. 5(1)(f))

What is prohibited: AI systems by government agencies that score citizens based on social behavior.

Example: Automated decisions about social benefits based on "risk scores".

Exceptions

Certain biometric applications are allowed:

  • Facial recognition for law enforcement (with authorization)
  • Emergency searches (missing persons, terrorism)
  • Medical applications
  • Critical infrastructure security

What Companies Must Do Now

  1. Check all AI systems for prohibited practices
  2. Update documentation of existing systems
  3. Review contracts with AI providers
  4. Create internal guidelines for permitted AI use

Sources

Related articles: EU AI Act · EU AI Act Checklist

For implementation support, find resources at ai-engineering.at.

Next step: operationalize compliance

Use ready-to-run GDPR templates, checklists and practical guidance for AI systems that need documentation and auditability.

Why AI Engineering
  • Local and self-hosted by default
  • Documented and auditable
  • Built from our own runtime
  • Made in Austria
Not legal advice.