EU Requires Digital Watermarks on Realistic AI-Generated Content Starting August 2
From August 2, the European Union will force companies to label AI-generated images, audio and text that appear authentic, using a digital watermark.
Under the EU’s AI Act, a labeling regime will take effect on August 2, obligating providers of AI-generated images, audio and text that mimic real media to embed a digital watermark. The rule targets new AI tools launched in the bloc, while older systems receive a four-month extension to adapt. Personal content such as private chats is excluded, as are works that are obviously artistic, satirical or fictional.
EU parliament member Sergey Lagodinsky emphasized that the measure protects both consumers and democratic discourse. Companies may use the EU-provided black-and-white label or create their own, and failure to comply could result in penalties of up to three percent of annual turnover. Industry voices note that many platforms already label AI media, but gaps remain, especially in advertising and publishing.
Why it matters
Labeling realistic AI content helps consumers spot synthetic media and safeguards the integrity of information online.
In this story