Starting August 2, new regulations have come into force across the entire European Union, fundamentally changing the landscape of digital interaction. According to Article 50 of the "AI Act," mandatory transparency is introduced: content publishers are required to label materials created with the help of artificial intelligence, as well as notify users when they are interacting with chatbots rather than real people.
Labeling Deepfakes and Generated Texts
The new regulations aim to protect the information ecosystem from disinformation. Now, any content in which AI participated in its creation must be marked with a special AI symbol. This applies to so-called "realistic-looking" deepfakes — images, audio, and video that imitate reality. If the material (be it music, a work of art, or a news summary) is created entirely by artificial intelligence, it must bear the label AI-Generated ("Created by AI").
Regulators have provided specific examples falling under these rules: deepfake videos featuring politicians, fictional events, AI-generated musical compositions, and texts. Furthermore, systems capable of recognizing user emotions or processing their biometric data are also subject to mandatory labeling.
Control over Developers and Accountability
The European Commission is placing special oversight on providers of general-purpose AI models. Organizations developing such systems are required to document their activities, publish summaries of training data, and implement a strict policy regarding copyright. National authorities of the participating countries will monitor compliance with these requirements.
"The goal is simple: whenever AI plays an important role, you have the right to know about it," explains the European Commission's position. The initiative is designed to help users make informed decisions and assess the level of trust in the information they receive.
Fines and Exemptions
The legislation provides for severe sanctions for violating transparency rules. Companies may be fined up to €15 million or 3% of their annual revenue. For EU institutions and agencies, the fine will be up to €750,000. However, the principle of proportionality will allow for mitigated penalties for small and medium-sized enterprises.
It is important to note that the law does not apply to personal content, such as messages in group chats, as well as "clearly artistic" satirical and fictional works.
Industry Reaction
The implementation of these rules makes Europe the world's most influential regulator of artificial intelligence technology. Industry representatives have already expressed concern, calling the new norms an excessive burden that could slow down innovation. However, European regulators are convinced that a single market with harmonized rules will, conversely, simplify development for AI providers by eliminating risks to health, safety, and fundamental citizens' rights.
