Microsoft Research Asia recently presented an artificial intelligence called VASA-1, which is able to generate highly realistic deepfake videos from a single photo and an audio clip.