BRUSSELS, BELGIUM / RankWire.AI / – European Union transparency standards for AI-produced content commenced across the bloc on August 2, 2026. Under Article 50 of the EU AI Act, entities are now mandated to provide clear notices for specific artificial intelligence interactions and synthetic media. These regulations apply to both developers creating AI systems and users deploying them to publish or display content. Importantly, a visible label on every AI-generated item is not required. These obligations are part of the phased rollout of the law following its approval in 2024.
Providers offering systems that directly engage with users must inform them when they are interacting with AI. Such a notice is not needed if the artificial origin of the interaction is obvious to a reasonably informed individual. Additionally, providers of systems that produce synthetic text, audio, images, or video must incorporate machine-readable indicators. These marks should enable the identification of AI-created or manipulated material, where technically feasible. Basic editing tools that do not significantly alter input or meaning remain outside this requirement.
The legislation differentiates these technical marking duties from the visible disclosures that deployers must provide to their audiences. Deployers are responsible for identifying deepfake images, audio, and video when the content might deceive viewers into believing it is authentic. They are also required to label AI-generated texts published for public interest information. However, this text labeling obligation does not apply after a human review or editorial oversight when an individual or organization assumes editorial responsibility. The information must be delivered in a clear, distinguishable, and accessible manner.
Distinction Between Machine Markings and Public Disclosures
Article 50 also mandates notices for emotion recognition and biometric categorization systems, which must be clearly displayed to individuals exposed. These disclosures should appear no later than the initial interaction or exposure. The law provides some exemption for clearly artistic, creative, satirical, or fictional works to protect their normal presentation and enjoyment. For deepfakes, disclosures in creative works may be limited but cannot be omitted entirely. Specific exceptions are also provided for certain law enforcement uses authorized by law and accompanied by safeguards.
The European Commission published comprehensive guidelines on July 20 to clarify the scope, definitions, and available exceptions. Furthermore, a voluntary Code of Practice was released on June 10 for providers and deployers of generative AI systems. Companies may adopt the code to demonstrate compliance with marking and labeling duties across EU member states. Non-membership in the code requires implementing other suitable measures and showing authorities that those measures meet legal requirements. Additionally, optional icons have been introduced, but these icons alone do not establish legal compliance.
Enforcement primarily rests with national authorities
Enforcement of the new transparency standards is mainly handled by national market surveillance agencies. The AI Office supervises limited cases involving systems linked to general-purpose AI models or large online platforms. The European Data Protection Supervisor oversees systems used by EU institutions, bodies, and agencies. Penalties for violations can reach up to 15 million euros or 3% of global annual turnover, with lower thresholds for smaller companies. Penalty procedures are determined by member states, which may employ warnings or other non-monetary enforcement actions.
A brief transition period applies for generative AI systems marketed before August 2, 2026. Providers of these existing systems have until December 2, 2026, to comply with the machine-readable marking requirement. This transition only addresses Article 50’s technical marking and detection obligations and does not delay other transparency responsibilities, which are already effective across the European Union. Content created before August 2 does not require retroactive labeling, but all new material must adhere to the relevant disclosure and marking rules.