AI Dark Patterns: Prohibited Practices Under Article 5 of the EU AI Act

While the labeling requirement (Art. 50) doesn't take effect until August 2026, the prohibitions under Article 5 of the EU AI Act have been in force since February 2, 2025. This means: Anyone using prohibited AI practices on their website today is already violating applicable EU law.

The problem: Many companies use third-party scripts that employ exactly these prohibited techniques – often without knowing it.

What Does Article 5 Specifically Prohibit?

The EU AI Act defines a clear list of unacceptable AI risks that are completely banned:

1. Subliminal Manipulation AI systems that use subliminal techniques to unconsciously influence people's behavior, causing them to make decisions they would not otherwise have made. Examples on websites:

• Dynamic pricing based on emotional state
• AI-controlled urgency timers creating artificial scarcity
• Personalized manipulation of display order to push purchases

2. Exploitation of Vulnerabilities AI systems that specifically exploit vulnerabilities of certain groups – whether due to age, disability, or economic situation. Examples:

• AI chatbots that steer elderly users toward expensive products
• Algorithms that bombard financially struggling users with credit offers

3. Workplace Emotion Recognition AI systems for emotion recognition in the workplace and educational institutions are prohibited (with few medical exceptions). This also affects web tools that analyze employee mood via webcam.

4. Social Scoring AI systems that rate and disadvantage people based on their social behavior. This also applies to customer scoring systems that discriminate disproportionately.

Which Third-Party Scripts Are Dangerous?

Many website operators unknowingly embed scripts that could fall under Article 5:

• Emotion detection APIs: Scripts analyzing user mood via webcam or mouse movement
• Manipulative A/B testing tools: When the algorithm exploits vulnerabilities
• Dark pattern plugins: Fake countdown timers, bogus "only X left" displays
• Persuasion AI: Chatbots employing psychological manipulation

The Penalties: Up to 35 Million Euros

Violations of Article 5 face the harshest penalties. The EU provides for fines of up to 35 million euros or 7% of global annual turnover – whichever is higher.

How to Protect Yourself: AIActify Scanner

The AIActify Risk Scanner analyzes your website for potential violations:

• Detection of emotion detection scripts
• Identification of manipulative AI patterns
• Review of embedded third-party scripts for Article 5 compliance
• Detailed risk report with concrete recommendations

Scan your website for free now and find out if you're already violating the EU AI Act today.

Start free scan