Abstract

Deepfakes, audio and video content generated or manipulated by artificial intelligence, pose a significant challenge for the security of information space. Identifying risks and threats on the level of the European Union contributed to the implementation of provisions directly referring to deepfakes in the Act on artificial intelligence (AI Act) adopted in 2024. They cover, i.a. a definition of deepfakes and requirements regarding transparency of synthetic media. The article points out faults in current legislative proposals and chances of regulating deepfakes, which require activity from public institutions. The article also presents a broader social and philosophical reflection on a comprehensive harmfulness of deepfakes and their negative influence on the perception of reality. The purpose of the article is to summarize the provisions of the AI Act regarding deepfakes, which could effectively protect individuals and a sensu lato public order against harmful use of deepfakes.