Transparency Requirements: Article 50 Obligations

Article 50 of the EU AI Act establishes transparency obligations for certain AI systems, even when they are not classified as high-risk.

Who Must Comply?

Transparency requirements apply to providers and deployers of specific AI systems regardless of their overall risk classification.

Key Obligations

1. Chatbots and Conversational AI

Users must be informed that they are interacting with an AI system, unless this is obvious from the circumstances.

2. Emotion Recognition Systems

Individuals must be informed when an emotion recognition system is being applied to them.

3. Biometric Categorization

People exposed to biometric categorization systems must be notified about the system's operation.

4. AI-Generated Content (Deepfakes)

Content that has been artificially generated or manipulated (text, audio, image, video) must be labeled as such in a machine-readable format. This includes:

• Deepfake videos: Must be clearly marked as AI-generated
• Synthetic audio: Must disclose AI generation
• AI-generated text: Must be labeled when published to inform on matters of public interest
• Manipulated images: Must disclose AI manipulation

Exceptions

Transparency labeling may be waived when:

• AI content is used in creative or artistic works (with proper credits)
• Law enforcement has legitimate authorization
• Content undergoes human editorial review

Implementation Steps

1. Identify all AI systems that generate or manipulate content
2. Implement disclosure mechanisms (labels, watermarks, metadata)
3. Ensure users can easily identify AI-generated content
4. Document your transparency measures

Assess Your Obligations

Use EUAIActReady to determine which transparency requirements apply to your AI systems.