BRUSSELS, BELGIUM / RankWire.AI / – European Union has enacted transparency regulations for artificial intelligence that became effective across the bloc on August 2, 2026. Under Article 50 of the EU AI Act, chatbots, deepfakes, synthetic media, and certain public-interest texts are now covered. The legislation distinguishes between technical marking responsibilities and notices visible to audiences. It does not mandate visible labels on every AI-created item. Instead, the obligations depend on the specific system, the type of content, and how an organization presents it.
Organizations providing interactive AI must disclose when they are engaging with users through artificial intelligence. They are exempt from this requirement if a knowledgeable user can readily recognize the system’s artificial nature. Additionally, generative AI providers are required to embed machine-readable signals into synthetic text, audio, images, and videos. These signals are intended to assist detection tools in identifying generated or manipulated material. When the current technology makes it feasible and reliable, providers must adopt effective marking methods.
Content publishers responsible for synthetic outputs face distinct disclosure requirements. They need to clearly identify deepfake images, recordings, and videos that might seem authentic. When it comes to AI-generated text on matters of public interest, they must label it in specific circumstances. However, human oversight and editorial review can eliminate the need for this particular labeling requirement. The responsible individual or organization must accept accountability for the published material before any exemption applies.
Visible notices aim to reveal deceptive synthetic media
Article 50 also covers emotion recognition and biometric classification systems used to evaluate individuals. Operators are required to inform affected persons when deploying these tools, with certain legal exceptions. The law provides more flexible disclosure rules for artistic, fictional, creative, and satirical works, ensuring notices do not disrupt the normal display or enjoyment of such content. Nevertheless, organizations must still provide appropriate disclosures if deepfake elements are involved.
European Commission has released guidance outlining the transparency rules, definitions, and exceptions available. It also launched a voluntary code for developers and organizations deploying generative AI, allowing participants to demonstrate how their marking and labeling practices satisfy legal standards. Companies not participating in the code must implement other effective compliance strategies. While optional icons can enhance public awareness, they alone do not confirm adherence to the EU AI Act.
Authorities are empowered to enforce compliance measures
Enforcement primarily falls on national market surveillance authorities within each EU member state. The European Data Protection Supervisor oversees relevant AI systems used by EU institutions and agencies. Violations of transparency rules may lead to fines of up to 15 million euros or 3% of a company’s global annual turnover. Smaller enterprises could face lower maximum penalties based on the lesser of these amounts.
A transition period applies to generative AI systems introduced before August 2, 2026. Providers of such systems have until December 2, 2026, to comply with the machine-readable marking obligation. This extension covers only the technical marking requirement and does not delay disclosures related to interactive AI, deepfake notices, or rules concerning public-interest texts. The legislation also does not mandate retroactive labeling for content created prior to the transparency rules coming into effect.