The Daily Star’s FREE newsletter is spectacular! Sign up today for the best stories straight to your inbox
Deepfake technology is creating fake human beings to spread misinformation, according to chilling reports.
Artificial intelligence-generated deepfake videos that can fool voters into thinking a political candidate has said or done something truly evil are already widespread.
Author and broadcaster Nina Schick, who has written extensively about the dangers of deepfakes, recently told Daily Star Online that 98% of deepfakes are porn clips featuring the faces of innocent people who may not even know their likenesses is being used.
But now new advances in AI are making humans unnecessary. Computer-generated faces, indistinguishable from real peoples’ photos, have been identified in social media campaigns from China, Russia and elsewhere.
The AI-generated faces are created using a technology known as generated adversarial networks, or GANs.
They consist of a pair of networks, one with the job of creating an imaginary face, the other which criticises the results again and again until the image is perfect.
These fake images can help sway public opinion by creating the impression of a whole crowd of people who all think the same way.
A website called This Person Does Not Exist publishes a near-infinite number of the fake faces online, created in real time every time you refresh the page.
“A year ago, this was a novelty,” says Ben Nimmo, director of investigations at social media intelligence group Graphika. “Now it feels like every operation we analyse tries this at least once.”
With GAN technology available across the web, it’s impossible to be certain that the stranger you’re talking to on Facebook or Twitter has ever existed.
Max Rizzuto, a research associate at the DFR Lab, believes that anyone can learn to spot a fake face: “Once you tell these people that it is actually a fabrication,” he told the Financial Times, “you can see this evolved sense that all humans have to spot abnormalities in an image.”
But Nina Schick isn’t so sure. She believes that the power of artificial intelligence to create convincing faces is advancing faster than our ability to spot them. Already, she says, only an AI can spot whether another AI has created an image.
Even Rizzuto accepts that it’s only a matter of time. “The potential capacity to deceive is kind of outweighed by the amount of labour that would take to pull it off [today],” he said. “In the near future, I would expect to see that to. . . diminish considerably.”
- Artificial Intelligence
Source: Read Full Article