Every year it becomes easier to create fakes. For example, today you can mount a video allegedly from a hidden camera, where a person confesses literally anything. Both the voice and the face will be indistinguishable from the real ones. Developments, able to create deepface, involved dozens of companies. Services are already unable to synthesize audio. Artificial intelligence analyzes audio recordings of a real person and simulates speech. The more material (audio recordings) for training the neural network, the more difficult it is to distinguish synthesized speech from the recording of a real person. There are also programs that superimpose the facial expressions of the managing actor on any other person, resulting in a deepfake video.
According to RT. Author Oleg Adamovich