AI Labelling Requirement

Under Article 50 of the EU AI Act, graduated transparency duties have applied across the EU since 2 August 2026: chatbots must make clear to users that they are interacting with an AI; AI-generated or AI-manipulated image, audio and video content that resembles existing persons, objects or events and would falsely appear authentic (deepfakes) must be disclosed by deployers; and AI-generated text published to inform the public on matters of public interest is subject to a disclosure duty as well – which falls away where the content has undergone human review and a natural or legal person holds editorial responsibility for its publication. For evidently artistic, creative, satirical or fictional work, the second subparagraph of Article 50(4) provides for a lighter form of disclosure. The split of roles matters here: providers of the systems have to mark outputs in a machine-readable format and make them detectable as artificially generated under Article 50(2), whereas deployers owe the disclosure that people actually see under Article 50(4); whether a company is a provider or a deployer in a given case depends on how it uses and passes on the system. For providers of systems generating synthetic content that were already placed on the market before 2 August 2026, a transitional deadline of 2 December 2026 applies for complying with Article 50(2). The EU AI Act Service Desk article page carries an amendment notice regarding the Digital Omnibus and still shows (as of 14 August 2026) the wording before that amendment – the consolidated version in the Official Journal is the authoritative one. Individual platforms may require their own, sometimes stricter, labels on top of that. Whatever the precise legal position, an open and transparent approach to AI assistance in image, audio, video and text content is advisable for companies, not least as a trust signal.