The rise of generative adversarial networks, a method of training content-faking algorithms, has an expert in fake videos concerned for the future of video as a source of truth. In addition to footage, provenance and authentication will be required to assure viewers they are seeing what the camera captured and not something added in later by a human or machine.
[Fakers] can create fake images or short videos using machine learning techniques: in particular, generative adversarial networks (GANs), which learn to generate fake content. These pit a network that generates fake content against a ‘classifer’ network that attempts to discriminate between real and fake content, so that the faking network rapidly improves.I’ve seen the technology get good enough that I’m now very concerned. In 5 or 10 years, this is going to get really good. At some point we will reach a stage where we can generate realistic video, with audio, of a world leader, and that’s going to be very disconcerting. I would say that the field of digital forensics is now behind in video.