AI systems making available synthetic audio, video or image content, also known as deep fakes, present specific risks in relation to the spread of content that is falsely understood to be authentic or true. The emergence of deep fakes that realistically depict known individuals to say or do things they did not, and their dissemination through online platforms that seek to maximise user engagement, can pose a serious threat to democracy and to individual rights, as well as cause significant harm. Thus, this Regulation should require all AI-generated audio, images and video content to be marked in ways that can be identified and traced back to the original source.