BRUSSELS, BELGIUM / RankWire.AI / – Effective on August 2, 2026, new European Union transparency regulations for artificial intelligence have been enacted. Article 50 of the EU AI Act now specifies how businesses must identify certain AI interactions and synthetic materials. The law encompasses chatbots, deepfakes, generated media, and some public-interest texts. It does not require visible labels on all AI-produced content. Instead, each obligation depends on the system, the output, and how an organisation presents it.
Operators of interactive AI systems are now required to inform users when they are engaging with artificial intelligence. This rule does not apply if a typical user can easily recognize the AI’s artificial nature. Providers of generative AI must also embed machine-readable information into synthetic text, images, audio, and video. This data should help automated detection of altered or generated material. The requirement is applicable when available technology can provide reliable and effective marking.
Entities distributing synthetic media have separate disclosure obligations. They must identify deepfake images, audio, or video that could seem genuine. They are also required to label AI-generated texts on public-interest topics in specific cases. However, this does not apply if the material is reviewed and editorial control is exercised. A person or organisation responsible for publication must accept accountability before the exemption applies.
Transparency rules focus on deepfakes and civic content
The regulations also cover emotion recognition and biometric categorization systems used to assess individuals. Operators must notify affected users when employing these technologies, except in limited legally authorized cases. The EU AI Act provides more flexibility for creative, fictional, artistic, and satirical works. Such works must maintain their usual presentation and enjoyment. Nonetheless, operators still need to disclose when a work includes deepfake content.
The European Commission issued guidance clarifying the scope of Article 50 and its exemptions. It also released a voluntary code for companies developing or using generative AI. Firms can use this code to demonstrate how their labeling and marking meet the legal standards. Those not participating must implement other effective compliance strategies. Optional icons can help raise public awareness, but these symbols alone do not ensure compliance.
Authorities are empowered to impose hefty fines
Enforcement will primarily be led by national market surveillance authorities within the EU. The AI Office will oversee a smaller set of systems connected to general-purpose AI models and major platforms. The European Data Protection Supervisor will monitor relevant systems used by EU institutions and agencies. Fines of up to 15 million euros can be issued for violations. The law also permits penalties equal to 3% of a company’s global annual turnover.
Existing generative AI systems on the market before August 2 have a limited transition period. Their providers have until December 2, 2026, to comply with the machine-readable marking requirement. This extension only applies to the technical aspect of detectable synthetic content. It does not delay chatbot notices, deepfake disclosures, or rules about public-interest texts. Content created before the transparency rules took effect does not need to be labeled.