Labelling deepfakes in terms of EU law: The AI Act and beyond

Generative artificial intelligence (GenAI) can create various types of content. Some of this content includes so-called deepfakes, which resemble actual content, such as the likeness of famous individuals, but are not genuine. The latest EU regulation – the Artificial Intelligence Act (AI Act) – introduces the requirement to label such content as artificially generated or manipulated. This article discusses these obligations and explains the provisions of the AI Act. It also outlines the practical implications of the new regulations and how to comply with them. However, the AI Act is not the only legal framework mandating the labelling of AI-generated content. The article’s thesis, which was ultimately confirmed, is that the AI Act is not the sole source of such obligations under EU law. Similar requirements also stem from EU competition and consumer protection laws, specifically Directive 2006/114 and the Unfair Commercial Practices Directive. The article provides a detailed analysis of these directives, which may serve as a basis for requiring market participants to label GenAI-generated content. It also explains how entities using AI can comply with these obligations, offering examples of specific actions. The article was prepared using the legal-dogmatic research method.