Article 50 is an independent transparency layer with four operational limbs: AI interaction disclosure, synthetic content marking and detection, emotion-recognition or biometric-categorisation notice, and deepfake or public-interest text disclosure.
The Four Article 50 Operational Limbs
- Article 50(1): Providers handle AI interaction disclosure for AI systems intended to interact directly with natural persons, unless obvious in context
- Article 50(2): Providers handle synthetic content marking and detection for synthetic audio, image, video, or text outputs
- Article 50(3): Deployers handle emotion-recognition and biometric-categorisation notice for exposed natural persons
- Article 50(4): Deployers handle deepfake and public-interest text disclosure with human-visible disclosure where required
Exemptions from Article 50
Article 50 contains paragraph-specific exceptions, including authorised criminal-law purposes, the Article 50(1) obviousness exception, Article 50(2) standard editing or no-substantial-alteration cases, and Article 50(4) human review/editorial responsibility for public-interest text. Article 50(5) is about clear, distinguishable, accessible delivery at first interaction or exposure.
Regulation (EU) 2024/1689 — Article 50(1)
“Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect.”