Compliance 4 min read

Deepfakes & Chatbots: Transparency Obligations Explained

Article 50 lays out specific transparency rules for AI systems that interact with humans or generate content. Here is what you need to label and disclose.

By Jessica Chen · Published

The "Right to Know"

One of the AI Act's central tenets is that humans should know when they are interacting with a machine. Article 50 codifies this "right to know" for specific types of AI.

Who Must Comply?

  • Chatbots & Customer Service AI: You must inform the user that they are communicating with an AI system, unless it's obvious from the context.
  • Emotion Recognition: If an AI is analyzing emotions or biometric data, the subject must be informed.
  • Deepfakes & Synthetic Media: AI-generated content (audio, video, image) that resembles existing persons, places, or events must be clearly labeled as artificially manipulated.

Exceptions

Exceptions exist for law enforcement and for authorized artistic/creative works, but the default rule is clear: Label it.

Implementation Tips

Don't bury the disclosure in a Terms of Service link. It needs to be prominent and timely—at the moment of interaction.