Deepfakes on the EU AI Act and Its Implementation in the Newsrooms

This article provides an analysis of the EU Artificial Intelligence Act, particularly on deepfakes, which regulates Artificial Intelligence in the field of journalism. Regulators dedicate an important section to deepfakes to protect fundamental rights and curb the spread of misinformation, which poses an open threat to democracy and the media ecosystem. We note that when using Artificial Intelligence systems and tools, media organizations and journalists must apply the provisions of the EU Artificial Intelligence law and be transparent with their audiences and at the same time, they have to apply their Code of Ethics. Nevertheless, they are obliged to avoid producing synthetic content that is inaccurate and/or misleading to the public. Synthetic, video, audio, images, and text should be watermarked, Nevertheless media and journalists are prohibited to produce synthetic content which might mislead the audience, even if it is watermarked. On the other hand, we have tested and found that Artificial Intelligence systems such as Chat-GPT don’t refrain from providing synthetic content (images and videos) of public figures with adequate prompting. Thus, the burden falls on the shoulders of the Journalists and the Media organisations.