ByteDance unveils OmniHuman-1 AI model for ultra-realistic video creation

ByteDance has launched a new AI model called OmniHuman-1, which can create ultra-realistic human videos from a single image. This development comes amid ongoing concerns about TikTok's influence in the U.S. and the potential misuse of deepfake technology. OmniHuman-1 has been trained on over 18,700 hours of video, allowing it to produce lifelike movements and speech. Unlike previous models that required multiple images, this one can generate convincing videos from just one, raising concerns about its use in disinformation and identity theft. The model's release coincides with rising global issues related to AI-generated disinformation, particularly in elections. As ByteDance advances in AI technology, the U.S. is working to enhance its own capabilities to address these emerging threats.