On 10 June 2026, the European Commission published the Code of Practice on Transparency of AI-generated content. It is the first document that gives creators and platforms concrete guidance on how to meet the transparency obligations of Article 50 of the AI Act. The duty to label synthetic content already applied, but there was no operational manual for it.
Signing up to the code is voluntary, and the Commission has opened it to a wide range of companies. Eligible signatories include “All providers and deployers of generative AI systems with existing or planned operations in the EU market”. The same framework therefore applies to providers of generative tools and to the in-house marketing teams of brands.
What the code covers
The document is divided into two parts according to the role in the production chain. The first targets providers of generative systems and sets rules for machine-readable marking and detection of content created or modified by AI, the obligation under Article 50(2). The second part targets the entities that deploy such content and covers the visible labelling of deepfakes and of text created or modified by AI under Article 50(4).
A practical aid is added in an annex. The Commission has prepared an optional EU icon in three variants that a deploying entity can rely on when it wants to meet the labelling obligation consistently across the market. The company does not have to devise its own visual solution.
Deadlines and the transitional period
The transparency obligations become binding on 2 August 2026. Systems placed on the market before that date have a deferral. The Commission grants them a transitional period until 2 December 2026, during which they have time to bring their labelling into line with the requirements.
The code is currently undergoing an assessment of compliance by the Commission and the AI Board, which has been running since its publication. Signing up to it works as a verifiable way of meeting the obligations, but it does not move the August deadline.
Penalties for non-compliance
Fines are set out in Article 99 of the AI Act. Breaches of the transparency obligations risk penalties under Article 99(4): “up to €15,000,000 or, if the offender is an undertaking, up to 3% of its total worldwide annual turnover for the preceding financial year, whichever is higher”. For a large European e-commerce company, the 3% cap on turnover is more painful than a fixed sum.
Operational impact on brands
Labelling of AI content is moving from principle to operations. Creators and brands now have ready-made labels and clearly described obligations, so the excuse of a missing standard no longer works. The gap between rule and practice remains, however. At its WWDC developer conference, Apple added generative photo editing to the iPhone and said nothing about visible labelling for users, even though edited images published in the EU fall under the same regime. Anyone in Europe using AI in content now has a clear deadline and a clear penalty ceiling.
Comments