EU requires labels for deepfakes and AI-generated content

4 min read

From this week the EU's AI Act requires providers and professional users of generative AI to make artificially generated or manipulated content clearly recognisable, with stricter duties for realistic deepfakes that imitate real people. The law builds on a voluntary code of practice that promotes machine-readable markings and visible labels, and it follows the creation of an EU AI Office. Yet experts and major AI companies warn that technical limits — fragile watermarks, lack of interoperable standards, and content created outside the EU — could hinder detection and enforcement, creating a gap between rules and real-world impact.

The AI Act mandates labelling of synthetic media starting this week.