Deepfake is something we will all become familiar with after a number of videos have hit social media’s TikTok app controlled by the Chinese Communist Party (CCP).
Deepfakes are created using Artificial Intelligence (AI) technology and are becoming very convincing and sophisticated.
Famous actor Tom Cruise has been depicted performing a magic trick, playing golf, and speaking about Mikheil Gorbachev. The concern is they all appear to be deepfake videos.
Although the account DeepTomCruise had just three videos posted, in one week, they have been seen by 11 million internet users, garnered 1.1 million likes, and collected 365,000 followers. These videos now appear to have been taken down from that account.
The videos not only look real but sound it as well. To the discerning eye, however, there are a few minor hiccups. Tom puts on his sunglasses in the golf shot video, but at one point, they disappear from the frame, something you can only pick up when the frame rate is slowed.
Deep fakes are getting scary good and taking over TikTok. Every public figure should just be on there with a verified account—even if they don’t want to make content—to make it easier to identify their fakes. Here’s Tom Cruise: pic.twitter.com/xoSJt1bvVR
— lauren white (@laurenmwhite) February 25, 2021
The magic trick video is compelling, right down to that maniacal laugh we all know.
“I’m gonna show you some magic … It’s the real thing,” he said while holding up the coin that he soon makes disappear. “It’s all the real thing,” he says with intensity, waving his hand over his head in a clear meta-joke about his fake face—and once again ending with a wild laugh, reported New York Post.
This is very much in the top 5% of deepfakes out there in terms of quality,” leading expert Henry Ajder told the UK Times.
“It would have taken many hours generating the face swap and applying post-production edits after an excellent performance by an impersonator,” he tweeted.
A system called Generative Adversarial Networks (GAN) was developed by Canadian doctoral student Ian Goodfellow, the director of Machine Learning at Apple. GAN employs two deep learning neural network AI’s in tandem. One network uses photos and videos of the real person, making a template to draw synthesized images. The second network separates the real thing from copies, which enhances the final product’s accuracy.
In 2018, a video was published on YouTube showing President Barack Obama saying, “We’re entering an era in which our enemies can make it look like anyone is saying anything at any point in time.” To illustrate its point, it continues, “They could have me say things like, I don’t know … how about this, simply: ‘President Trump is a total and complete [expletive removed].’”
The video then pans to a dual-screen of actor Jordan Peele doing the talking, demonstrating how the technology works in real-time.
Rachel Tobac, the CEO of online security company SocialProof, tweeted that the videos illustrate we have reached a stage of almost “undetectable Deepfakes.”
“Just because you feel you can personally tell the difference between synthetic & authentic media, it doesn’t mean we’re good to go. It matters what the general public believes,” she wrote.
2 years ago on stage I was asked “when will Deepfake video/audio impact trust & be believable in social engineering?” My response then was that we were 2 years away from undetectable Deepfakes. I wish my prediction then was wrong. We need synthetic media detection + labels ASAP. pic.twitter.com/yUUOTDepYY
— Rachel Tobac (@RachelTobac) February 26, 2021
“Deepfakes will impact public trust, provide cover & plausible deniability for criminals/abusers caught on video or audio, and will be (and are) used to manipulate, humiliate, & hurt people,” she said, adding they had “real-world safety, political, etc. impact for everyone.”
So, where is this deepfakery leading us? Bad actors will likely use it for purposes other than innocent fun.
“Mission Impossible” may have been accomplished.