BRUSSELS, BELGIUM / RankWire.AI / – European Union rules requiring transparency for some AI-generated content took effect across the bloc on August 2, 2026. Article 50 of the EU AI Act now sets duties for developers and users of artificial intelligence systems. The rules cover chatbots, synthetic media, deepfakes and certain public-interest text. They do not require a visible warning on every item created with AI. Instead, the law applies different requirements based on the system, content and method of publication.
AI systems that interact directly with people must disclose that users are dealing with artificial intelligence. The duty does not apply when the system’s nature remains obvious to an informed and attentive person. Providers of generative AI must also add machine-readable markers to synthetic text, audio, images and video. These markers should help detection tools identify generated or manipulated material. The requirement applies when the technology allows reliable marking without making the output unusable or fundamentally changing its normal operation.
The law places separate disclosure duties on organisations that publish or present synthetic material. They must identify deepfake images, recordings and videos that could appear authentic. They must also label AI-generated text that informs the public about matters of public interest. That text requirement does not apply after meaningful human review and editorial control. A person or organisation must also accept responsibility for publication. Required notices must remain clear, accessible and visible during the first interaction or exposure.
Deepfakes face direct disclosure requirements
Article 50 also covers emotion recognition and biometric categorisation tools in situations involving individuals. Operators must notify people when those systems analyse them, subject to limited legal exceptions. Creative, artistic, satirical and fictional works receive narrower disclosure treatment under the regulation. Notices for such works must not disrupt their normal display or enjoyment. However, the deepfake disclosure duty still applies in an appropriate form. Certain law-enforcement uses may qualify for specific exemptions when authorised by law and governed by safeguards.
The European Commission has published guidance and a voluntary code to support consistent compliance across member states. Providers and deployers may use the code to show how their marking and labelling systems meet legal requirements. Participation remains voluntary, but the transparency obligations themselves remain mandatory. Companies outside the code must use other effective measures and explain them to regulators. Optional icons may help audiences recognise synthetic content, although using an icon alone does not prove full compliance with the AI Act.
National regulators lead enforcement across the bloc
National market surveillance authorities hold primary responsibility for enforcing the transparency rules. The AI Office supervises limited cases involving certain systems linked to general-purpose AI models and major online platforms. The European Data Protection Supervisor oversees systems used by EU institutions and agencies. Breaches can lead to fines reaching 15 million euros or 3% of global annual turnover. Smaller companies receive lower applicable ceilings under the regulation. Member states may also use warnings and other corrective measures through their national enforcement systems.
A limited transition applies to generative AI systems that entered the market before August 2, 2026. Providers of those systems have until December 2, 2026, to meet the machine-readable marking requirement. The extension covers only that technical obligation. It does not postpone chatbot notices, deepfake disclosures or the rules for public-interest text. Content created before August 2 does not require retroactive labelling. New content and new interactions must follow the applicable transparency duties from the date the rules began.