The European Commission has finalised its Code of Practice on marking and labelling AI‑generated content, ahead of transparency rules under the AI Act coming into force on 2 August 2026. The voluntary code, developed with industry, civil‑society and Member State experts, is designed to show providers and deployers how to meet Article 50 obligations to mark AI‑generated or manipulated text, images, audio and video in machine‑readable ways.
The code is structured in two parts: one for providers of generative AI systems, and one for deployers who use those systems to publish content that informs the public on matters of public interest. Providers are expected to implement technical markings such as metadata, watermarks or fingerprinting and to offer detection tools so platforms and third parties can check whether content came from their models. Deployers must clearly label deepfakes and AI‑generated or manipulated news‑style text, with more flexible treatment for artistic or satirical work so labels do not disrupt how it is consumed.
For Scottish companies serving EU users, the message is simple: if you publish AI‑generated content into the EU, you are caught by these transparency rules even post‑Brexit. UK media, SaaS vendors and AI startups will need to align technical markings and front‑end labels with the EU standard or risk regulatory friction once the AI Act transparency obligations start to bite.
