Starting Sunday, all companies distributing AI-generated content within the European Union must label it clearly, under a new rule from the EU's AI Act. The mandate aims to help the public identify synthetic materials, from deepfake videos to AI-written articles, addressing growing concerns about misinformation. The requirement applies to platforms and content creators alike, with fines for non-compliance reaching up to 6% of global annual revenue for the largest firms.
The rule, part of a broader regulatory framework, follows years of debate over the risks of AI-generated media. It covers a wide range of outputs, including text, images, audio, and video, but exempts content that is purely artistic or satirical if clearly presented as such. Companies must use standardised metadata or visible markers, though the exact format remains flexible.
For tech giants like Google, Meta, and OpenAI, compliance means adjusting existing content moderation pipelines and updating user interfaces. Smaller startups worry about the administrative burden, while digital rights groups have raised concerns about potential overreach. The European Commission has promised guidance documents to clarify edge cases.
The move has been welcomed by fact-checkers and journalists who see it as a necessary step to preserve trust in public information. However, some critics argue the labeling could be easily circumvented and that the real challenge lies in enforcement across international borders.
As the rollout begins, watch for early high-profile fines or court challenges. The EU is also expected to publish a list of best practices for labeling within weeks. For consumers, the immediate impact may be subtle: look for a small icon or text notice on social media posts and news sites.