EU issues voluntary code for AI content labelling

The European Commission published a voluntary Code of Practice on June 10 to guide companies on labelling AI-generated and AI-manipulated content before mandatory rules in Article 50 of the EU AI Act take effect on Aug. 2, 2026. The Code sets out practical steps for firms that build generative models and for those that deploy them in user-facing products.

From Aug. 2, two types of content must be clearly flagged. Deepfakes and AI-generated or AI-manipulated text on matters of public interest must carry a label. Anyone interacting with a conversational AI system must be told they are dealing with a machine. The Commission says the requirement is intended to help people recognise AI-made or altered material and reduce the risk of deception in public debate.

The Code is voluntary; signing it gives a company a recognised way to show compliance but does not change legal obligations, which apply regardless. The guidance asks model builders to embed machine-readable markers in model outputs so downstream services can detect AI-produced content. Firms that deploy models must present visible labels to end users. For public-interest AI text, visible labelling is required when content is published without human review or editorial oversight.

The document recommends open technical standards and proposes a common EU icon for a consistent visual cue across services. The Commission opened the Code for signatures and encouraged providers and deployers to sign. The Code was drafted by six independent experts with input from more than 180 stakeholders and is the first instrument to address content labelling under the AI Act.

The Commission and the EU’s AI Board must still assess the Code’s adequacy. The executive arm plans to publish separate guidelines to clarify legal details and fill gaps left by the Code. Companies serving European users have fewer than two months to identify what they must label, implement the required technical changes and decide whether to sign the voluntary Code.

Henna Virkkunen, the Commission’s executive vice-president for tech sovereignty, security and democracy, said Europeans have a right to know whether what they see, hear or read has been made or altered by AI, especially when such content can shape public debate.