By feeding GANs and autoencoders large volumes of video footage and photos, it is possible to create digital representations of targets individuals wish to impersonate.