On May 8, 2026, the European Commission published draft guidelines explaining how to comply with the transparency obligations in Article 50 of the EU AI Act. These are the first official Commission instrument to interpret Article 50 across its full scope, and they are meant to apply alongside the law itself once those obligations take effect on August 2, 2026. The text was issued by the AI Office and opened for a targeted consultation that ran until June 3, 2026.
Article 50 sets the AI Act’s disclosure rules. Providers must tell people when they are interacting with an AI system, such as a chatbot, and must mark machine-generated audio, image, video, or text in a machine-readable way so that synthetic content can be detected. Deployers face their own duties: they must disclose deepfakes and AI-generated material on matters of public interest, and must inform people subject to emotion-recognition or biometric-categorization systems. The guidelines clarify the scope of these requirements and how they interact with the Commission’s separate draft Code of Practice on labeling AI-generated content.
Because the guidelines are non-binding interpretive guidance rather than new law, they do not change the underlying obligations, but they signal how the Commission and national authorities expect to read and enforce them. For any business that ships a chatbot, an image or text generator, or a deepfake-capable tool into the EU market, this draft is the clearest signal yet of what “transparency” will mean in practice when the rules become enforceable in August 2026.