BRUSSELS, BELGIUM / RankWire.AI / – European Union transparency rules for AI-generated content took effect across the bloc on August 2, 2026. Article 50 of the EU AI Act now requires clear notices for certain artificial intelligence interactions and synthetic media. The rules cover providers that build AI systems and deployers that use them to publish or present content. They do not require a visible label on every item created with AI. The obligations form part of the law’s phased implementation following its adoption in 2024.
Providers of systems that interact directly with people must tell users when they are dealing with AI. The notice is unnecessary when the artificial nature of the interaction is obvious to a reasonably informed person. Providers of systems that generate synthetic text, audio, images or video must also add machine-readable marks. Those marks must allow detection of AI-generated or manipulated material, as far as technically feasible. Standard editing tools remain outside this duty when they do not substantially change the input or its meaning.
The law separates those technical marking duties from the visible disclosures that deployers must give to audiences. Deployers must identify deepfake images, audio and video when the material could falsely appear authentic. They must also label AI-generated text published to inform the public on matters of public interest. The text rule does not apply after human review or editorial control when a person or organisation accepts editorial responsibility. People must receive the required information in a clear, distinguishable and accessible form.
Rules distinguish machine marking from public labels
Article 50 also covers emotion recognition and biometric categorisation systems, which require notices to exposed individuals. Disclosures must appear clearly and distinctly no later than the first interaction or first exposure. The law limits some duties for clearly artistic, creative, satirical or fictional works to protect normal display and enjoyment. For deepfakes, the disclosure can be limited in creative works but cannot disappear entirely. The regulation also provides targeted exceptions for certain law-enforcement uses authorised by law and subject to safeguards.
The European Commission issued detailed guidelines on July 20 to explain the scope, definitions and available exceptions. It also published a voluntary Code of Practice on June 10 for providers and deployers of generative AI systems. Companies may use the code to demonstrate compliance with marking and labelling duties across EU member states. Non-signatories must use other adequate measures and show authorities how those measures satisfy the law. The Commission also released optional icons, but the icons alone do not establish legal compliance.
Enforcement rests mainly with national authorities
National market surveillance authorities carry primary responsibility for enforcing the new transparency requirements. The AI Office oversees limited cases involving systems linked to general-purpose AI models or certain large online platforms. The European Data Protection Supervisor handles systems used by EU institutions, bodies and agencies. Violations can bring fines of up to 15 million euros or 3% of worldwide annual turnover, with lower ceilings for smaller businesses. Member states set penalty procedures and may use warnings or other non-monetary enforcement measures.
A narrow transition applies to generative AI systems placed on the market before August 2, 2026. Providers of those existing systems have until December 2, 2026, to meet the machine-readable marking requirement. The transition covers only Article 50’s technical marking and detection obligation. It does not delay the other transparency duties, which already apply across the European Union. Content generated before August 2 does not require retroactive labelling, while new material must follow the applicable disclosure and marking rules.