Article 50 of the EU AI Act requires that users are clearly informed when they interact with an AI system. This applies to chatbots, AI-generated content (text, images, audio, video) and emotion recognition systems. The goal is to prevent deception — people have the right to know when they are talking to a machine.
How to implement this:
- Chatbots and virtual assistants: Add a clear label such as "You are chatting with an AI assistant" at the start of each conversation. Keep it visible, not hidden in terms of service.
- AI-generated content: Label synthetic text, images or audio with a visible marker. For deepfakes or synthetic media, machine-readable metadata is also required.
- In-product notices: Add an "AI-powered" badge next to features that use AI to generate recommendations, summaries or decisions.
- Documentation: Record where in your product AI is used and how users are informed. This evidence is needed for compliance audits.
Transparency builds trust — users who know they are interacting with AI tend to set appropriate expectations and report issues more constructively. Tidal Control helps you map all AI touchpoints in your product and track transparency measures as auditable controls.