BRUSSELS, BELGIUM / RankWire.AI / – European Union rules to ensure transparency in artificial intelligence took effect across the bloc on August 2, 2026. Under Article 50 of the EU AI Act, the legislation now encompasses chatbots, deepfakes, synthetic media, and certain public-interest texts. The law distinguishes between technical marking obligations and notices visible to audiences, but does not mandate visible labels for all AI-created content. Instead, requirements vary depending on the system, the content, and the manner of presentation by organizations.
Providers of interactive AI must inform users when they are engaging with artificial intelligence. An exception exists when the user can distinctly recognize the system’s artificial nature without a notice. Generative AI providers are also required to embed machine-readable signals in synthetic text, audio, images, and video, aiding detection tools in identifying generated or modified content. When the technology permits, providers are obliged to employ effective marking methods that are technically feasible and reliable.
Organizations publishing synthetic content are subject to a different set of disclosure rules. They must label deepfake images, recordings, and videos that may appear authentic and, in certain circumstances, identify AI-generated text related to public interest topics. Human oversight and editorial control can exempt this specific text labeling requirement, provided that a person or organization accepts responsibility for the published material before such an exemption applies.
Visible disclosures aim to combat deceptive synthetic media
Article 50 also covers systems involved in emotion recognition and biometric categorization that evaluate individuals. Operators are required to notify affected persons when these tools are used, with limited legal exceptions. Artistic, fictional, creative, and satirical works are granted more flexible disclosure provisions, as notices must not interfere with their normal display or enjoyment. Nonetheless, organizations still need to provide appropriate disclosure when such content contains deepfake elements.
European Commission has published guidance outlining the transparency rules, definitions, and available exemptions. Additionally, it introduced a voluntary code for developers and organizations deploying generative AI. Companies participating in the code can showcase how their marking and labeling approaches align with legal standards. Those outside the framework must establish other effective compliance strategies. While optional icons can help raise public awareness, the symbols alone do not serve as proof of adherence to the EU AI Act.
Regulators acquire enhanced enforcement authority
Enforcement within EU member states primarily rests with national market surveillance authorities. The European Data Protection Supervisor oversees relevant AI systems employed by EU institutions and agencies. Violations of the transparency requirements can attract fines of up to 15 million euros, or 3% of a company’s global annual turnover. Smaller businesses may face lower maximum penalties if the calculation results in a lesser amount.
A transitional period applies to generative AI systems introduced before August 2, 2026. Providers of these systems have until December 2, 2026, to comply with the machine-readable marking obligation. This extension is limited to the technical marking requirement and does not delay disclosures related to interactive AI, deepfake identification, or content concerning public interest. The regulation also clarifies that labels do not need to be retroactively applied to content created before the transparency rules became effective.