They may look real but they’re fake: A rise in deepfake technology

NBC’s hit show “America’s Got Talent” concluded its 17th season in summer 2022. Alongside the typical lineup of singers, dancers and comedians, the fourthplace winner showcased a new talent all together. The group, named Metaphysic, focused on the use of deepfake technology. In their first audition, they edited judge Simon Cowell’s face onto a singer that was onstage.  Projecting the video onto the auditorium’s screen made it appear like Cowell himself was auditioning for the show. While wildly entertaining for fans, the rapid growth of this technology has raised concerns for many around the world.

History

Deepfake is an extension of the technology used in photoshop or face swap filters on Snapchat. The term “deepfake” originated on Reddit in 2017 by a user who posted sexually explicit videos with faces of celebrities such as Taylor Swift or Scarlet Johannson edited in. The technology allows users to make video and audio samples to create the illusion that a certain person said or did something that never happened. Deepfake technology is currently capable of editing an existing face onto a speaker in a current video, creating content with AI generated faces, and recreating someone’s face and voice. 

Accessibility

The general process of deepfake could be completed by anyone who has the correct software– a high-speed computer would be preferable to minimize processing time. However, websites and apps have also been available for public use and have provided ways for consumers to create their own versions of deepfake. Companies also could be hired to create deepfake media for clients.  

Process

The Guardian provides details on the general process. First, hundreds or thousands of images of two faces are processed using an encoder–a type of AI that identifies what is similar between the faces. Then, a decoder is used, separating the qualities of the faces. Once both of these steps are done, the algorithms are ready to combine the faces into one piece of media. The Guardian explains that “a compressed image of person A’s face is fed into the decoder trained on person B. The decoder then reconstructs the face of person B with the expressions and orientation of face A.”

Positive Effects

Metaphysic’s feature on “America’s Got Talent” showed that deepfake technology can certainly be used for entertainment. The act was well-received, earning a standing ovation from audience and judges after the initial audition. Judge Heidi Klum even said that “…it was the best thing […she’d] seen all day.” According to the Washington Post, this group hoped to market itself in the entertainment industries, creating believable replicas of late actors for movies. Metaphysic has also released their own platform called everyany.one.  Their website has allowed any user to create deepfaked, artificial icons of themselves starting with a single image that can then be processed by their artificial intelligence. This technology has allowed anyone to access this digital technique. Beyond entertainment, CNBC points out how deepfake could be used positively in numerous other professions such as in education or healthcare.  

Criticism from the Public

Many criticize the prevalence of this technology, its accessibility and even the way it was publicized on America’s Got Talent. Past patterns such as the Reddit publications of 2017 show that deepfake platforms have the potential to be used for harm. 

The MIT Sloan School of Management offered that deepfake was a threat to corporations and businesses. They cited a situation in March 2019 where a deepfaked voice, programmed to mimic the head of a U.K. energy firm, was able to request a transfer of money. This example of deepfake was allegedly employed by thieves, who used the technology to carry out their crime. 

Deepfake could also be used in a political setting, making believable videos where politicians appear to make claims or commit actions that never happened. This technology could have the potential to influence public perception of politicians and damage reputations. For this reason, Facebook banned deepfake videos from their platform prior to the 2020 election in the USA.

Additionally, deepfake influences the journalism industry by lowering trust in the media. Since any professional video could be manufactured artificially, the public may begin to assume all videos are fake and manufactured and disregard credible videos. 

In China

In late 2022, two videos of deepfaked, fabricated newscasters were published online by pro-China bots. The New York Times reported that these videos were “intended to promote the interests of the Chinese Communist Party.” Small details such as newscasters’ hair and slightly unsynced mouth movements led to the discovery of the manufactured videos In January, the Cyberspace Administration of China introduced laws to control the use of deepfake. These laws included identifying any content as modified, obtaining consent before using someone’s face, and abstaining from spreading fake news.  

Moving Forward

Even with an entertaining deepfake presentation on “America’s Got Talent”, its availability and potential to influence reputations, politics and many other fields proves concerning to many. At Hope College, our mission is “to educate students for lives of leadership and service in a global society.” Knowledge about deepfake is integral as our global society continues to advance.