Making an AI-generated fake of a video or audio clip is now a lot easier. Researchers at Samsung's AI Centre have devised a method to train a model to animate with an extremely limited dataset: just a single photo and generate the desired output. The researchers were able to achieve this effect by training its algorithm on facial features like- the general shape of the face, eyes, mouth shape, and more scraped from a public repository of 7,000 images of celebrities gathered from YouTube. As the team proves, its model even works on the Mona Lisa, and other single-photo still portraits.
In the video, famous portraits of Albert Einstein, Fyodor Dostoyevsky, and Marilyn Monroe come to life as if they were Live Photos. The new paper by Samsung's Moscow-based researchers, however, shows that using only a single image of a person's face, a video can be generated of that face turning, speaking and making ordinary expressions — with convincing, though far from flawless, fidelity. As researchers continue to come up with low-lift methods for making high-quality fakes, there's a concern that they'll be used against people in the form of propaganda — or to depict people in situations they'd object to, like pornographic videos.