01
Key points
- EU rules taking effect Sunday require clear labels for deepfakes and other AI-generated images, text, video and audio.
- Chatbot providers must disclose AI interactions through a noticeable opening message or a persistent on-screen label.
- AI-generated public-interest text requires labelling when produced without human editorial oversight; purely personal use is excluded.
- Existing AI systems have until 2 December to comply, while artistic, creative, satirical and fictional work is exempt.
- Companies that fail to meet the requirements face large fines, according to the EU rules.
How the story developed
A living timeline
All times shown in UTC
Latest development
From Sunday, new EU AI Act obligations will require AI companies to disclose the artificial origin of deepfakes and other generated content.
Latest development
EU transparency obligations for deepfakes, AI-generated content and chatbot interactions are due to take effect Sunday.
Source transparency