BRUSSELS, BELGIUM / RankWire.AI / – European Union introduced transparency regulations for AI-generated material that took effect across the bloc on August 2, 2026. Under Article 50 of the EU AI Act, clear notices are now mandated for specific artificial intelligence interactions and synthetic media. These rules target providers responsible for developing AI systems and deployers who utilize such systems to publish or display content. Not every AI-produced item needs a visible label; instead, these requirements are part of a phased rollout following the law’s adoption in 2024.
Providers directly engaging with users must inform them when they are interacting with AI. When the artificial nature of the interaction is obvious to a reasonably informed individual, such notices are not necessary. Additionally, providers of systems that generate synthetic text, audio, images, or video are required to add machine-readable indicators. These marks should enable the detection of AI-produced or manipulated content where technically feasible. Standard editing tools are exempt from this obligation if they do not substantially alter the input or its meaning.
This law delineates between the technical marking duties and the visible disclosures that deployers must provide to the audience. Deployers are required to identify deepfake images, audio, and video, especially when such material could be mistaken for authentic. They must also label AI-generated text shared publicly on matters of public interest. The requirement for labeling does not apply after a human review or editorial oversight when the responsible person or organization assumes editorial responsibility. All individuals must receive this information in a clear, distinguishable, and accessible manner.
Clarification on machine marking versus public disclosure
Article 50 additionally addresses emotion recognition and biometric categorization systems, which must provide notices to the individuals affected. These disclosures should appear clearly and distinctly no later than the first interaction or initial exposure. The law offers some exemptions for artistic, creative, satirical, or fictional works to safeguard their normal presentation and enjoyment. While disclosures for deepfakes in creative contexts can be limited, they cannot be entirely omitted. Certain exceptions are also outlined for law enforcement uses authorized by law and subject to safeguards.
On July 20, the European Commission issued comprehensive guidelines clarifying scope, definitions, and exceptions. It also published a voluntary Code of Practice on June 10 for AI system providers and deployers. Companies can adopt this code to demonstrate compliance with labeling and marking duties across the EU. Non-participants must implement alternative measures and demonstrate to authorities how those measures fulfill legal requirements. The Commission also released optional icons, though their use alone does not guarantee legal compliance.
Enforcement primarily entrusted to national agencies
The main responsibility for enforcing these transparency rules lies with national market surveillance authorities. The European Data Protection Supervisor oversees cases involving systems linked to general-purpose AI models or large online platforms, while the European Commission’s AI Office handles certain limited situations. Violations can result in fines up to 15 million euros or 3% of global annual turnover, with lower caps for smaller enterprises. Penalty procedures are established by member states, which may also resort to warnings or other non-monetary enforcement actions.
Existing generative AI systems introduced before August 2, 2026, are subject to a narrow transition period. Providers have until December 2, 2026, to comply with the machine-readable marking requirement. This transition only concerns the technical marking and detection obligations specified in Article 50. Other transparency duties already apply across the EU without delay. Content produced before August 2 does not need retroactive labeling, but all new material must adhere to the applicable disclosure and marking standards.