Analysis finds at least 150 channels on messaging app that are distributing AI-generated images and video
Millions of people around the world are creating and sharing deepfake nudes on the secure messaging app Telegram, a Guardian analysis has shown, as the spread of advanced AI tools industrialises the online abuse of women.
The Guardian has identified at least 150 Telegram channels – large encrypted group chats popular for their secure communication – that appear to have users in many countries, from the UK to Brazil, China to Nigeria, Russia to India. Some of them offer “nudified” photos or videos for a fee: users can upload a photo of any woman, and AI will produce a video of that woman performing sexual acts. Many more offer a feed of images – of celebrities, social media influencers and ordinary women – made nude or made to perform sexual acts by AI. Followers are also using the channels to share tips on available deepfake tools.
While there have long been Telegram channels dedicated to distributing non-consensual nude images of women, the widespread availability of AI tools means anyone can instantly become the subject of graphic sexual content viewable by millions.
On a Russian-language Telegram channel advertising deepfake “blogger leaks” and “celebrity leaks”, a post about an AI nudification Telegram bot promised “a neural network that doesn’t know the word ‘no’”.
