A chatbot that sounds human, an image generator that produces photorealistic content, a voice cloning tool - they all make it harder to distinguish machine-produced content from human-created work. The risks of sophisticated generative and interactive AI - misinformation, manipulation at scale, impersonation, and fraud - are growing. From 2 August 2026, the EU AI Act requires businesses to be open about AI: to tell people when they are talking to a machine, mark synthetic content so its origin can be detected, label deepfakes, and warn individuals when AI is analysing their emotions or categorising them biometrically. This briefing explains how the transparency obligations fall on providers and deployers of AI.