EU to require labels on realistic AI-made content

Artificially generated images, audio and text designed to look authentic must be labelled under the EU’s AI Act. Companies must make clear when people interact with an AI-powered chatbot or view manipulated media or AI-written material, with visible marks and digital watermarks for synthetic content designed to look truthful.

Text on matters of public interest must also be identified as AI-made when there is no human editorial oversight. The requirements cover new AI systems on the EU market from 2 August, while existing ones have an extra four months to comply. Personal content by users is excluded, and exemptions apply for evidently artistic, satirical and fictional works.

EU lawmaker Sergey Lagodinsky framed the measures as protecting customers and democracy by preserving the authenticity of facts online. Tech industry representatives say they back clearer deepfake disclosure but fear EU guidance defines the category too broadly, creating label fatigue similar to cookie banners across advertising, film and publishing.

Major platforms have already adopted some disclosure tools. TikTok says it requires creators to label realistic AI-generated media and has helped label more than 3bn pieces of content, while Google says its SynthID watermark has been applied to more than 100bn images and 60,000 years’ worth of audio.