BRUSSELS, BELGIUM / RankWire.AI / – Effective from August 2, 2026, the European Union has implemented new transparency regulations for artificial intelligence. Article 50 of the EU AI Act specifies how companies must identify certain AI interactions and synthetic content. The legislation addresses chatbots, deepfakes, generated media, and some public-interest texts. However, it does not require visible labels on all content created using artificial intelligence. Instead, the obligations vary depending on the system, the output, and how organizations present it.
Operators of interactive AI systems are required to disclose when they are communicating with artificial intelligence. This requirement does not apply if an average user can easily recognize the system’s artificial origin. Providers of generative AI must also embed machine-readable information into synthetic images, audio, videos, and text. Such information supports automated detection of manipulated or generated content. The measure should be implemented only when current technology can reliably provide accurate markings.
Organizations involved in distributing synthetic media are subject to separate public reporting rules. They must clearly identify deepfake images, audio, or videos that could seem authentic. Additionally, AI-generated texts related to public-interest issues must be labeled in specific situations. These labeling obligations do not apply if individuals review the material and exercise editorial control. Before this exemption applies, a named individual or organization must accept responsibility for the publication.
Focus on Public Disclosures for Deepfakes and Civic Content
The rules also extend to emotion recognition and biometric categorization systems used to evaluate individuals. Operators must notify those affected when using these technologies, except in limited cases authorized by law. The EU AI Act offers more flexible disclosure options for creative, fictional, artistic, and satirical works. These notices must not interfere with the usual presentation or enjoyment of such works. Nonetheless, operators must still provide an appropriate disclosure if a work contains deepfake material.
The European Commission has issued guidance to clarify the scope of Article 50 and its exceptions. It also released a voluntary code of conduct for companies developing or deploying generative AI systems. Companies can use this code to demonstrate how their labeling and marking processes comply with legal standards. Those who choose not to participate must rely on other effective compliance measures. While optional icons may help raise public awareness, they do not in themselves establish legal compliance.
Authorities Can Enforce Heavy Fines for Non-Compliance
Enforcement will be led by national market surveillance authorities across EU countries. The AI Office will oversee a select group of systems connected to general-purpose AI models and major platforms. The European Data Protection Supervisor will monitor systems used by EU institutions and agencies. Violators can face fines up to 15 million euros or penalties equivalent to 3% of their global annual turnover.
Generative AI systems already available before August 2 are granted a limited transition period. Providers have until December 2, 2026, to comply with the machine-readable marking requirement. This extension only applies to the technical aspect of marking synthetic content. It does not delay notices for chatbots, disclosures for deepfakes, or rules concerning public-interest texts. Additionally, organizations are not required to label content created before the transparency rules went into effect.