Egor Zakharov has created a new artificial intelligence system that can generate a fake video clip from a single photo. Deepfakes or fake clips that make the subjects say and do things they haven’t usually require huge datasets. This has been one of the factors keeping the technology away from bad actors.
However, the new software is cause for alarm. The development of the software by Samsung’s artificial intelligence lab in Russia has been detailed in a paper on the site arxiv. The paper's author says their research has resulted in a system that first performs ‘ lengthy meta-learning on a large dataset of videos, and after that is able to frame few- and one-shot learning of neural talking head models of previously unseen people as adversarial training problems with high capacity generators and discriminators.
Good and bad applications
Crucially, the system is able to initialize the parameters of both the generator and the discriminator in a person-specific way, so that training can be based on just a few images and done quickly, despite the need to tune tens of millions of parameters.
We show that such an approach is able to learn highly realistic and personalized talking head models of new people and even portrait paintings. The software can have fun applications like being portraits to life such as the example given in the video below of the Mona Lisa.
Fake news will rise
But what many fear is this software will end up in the wrong hands and be used as a tool for political and social manipulation. While most of us are still struggling to understand and monitor fake news in our feeds, deepfakes are next level. Some of the example shown in the video made from even just single images could fool many people, particularly when presented within a supposedly trustworthy context.
A video was heavily shared in Belgium last year that depicted Republic President Donald Trump saying: “As you know, I had the balls to withdraw from the Paris climate agreement,” he said, looking directly into the camera, “and so should you.” The video caused uproar and quickly went viral attracting multitudes of comments from outraged Belgians. Except the video was fake.
It had been created by a Belgian political party, Socialistische Partij Anders, or sp.a and shared through social media. The party had commissioned a local production company to create the closing deepfake software and had thought the poor quality of the clip would be clear to all viewers. Except they wrong.
Watching the clip knowing its a fake it is evident that is is not real, but it does seem to have fooled many who poured out their anger that the U.S was messing with the European country’s climate policy. What the story shows is that even a poor version of deep fake videos can cause distress. It isn’t clear what Samsung intends to use the recently created algorithm for but safe to say a new era of fake news is on the way.