Musk Pushes an Unlabeled Deepfake of Harris on X
The social media platform’s owner defends X as a bastion of free speech. But by not labeling an AI-faked video he’s violating the safety rules of his own company. It’s a great lesson in how not to be a leader.
Vice President Kamala Harris and Elon Musk.. Illustration: Inc; Photo: Getty Images
On Friday billionaire entrepreneur and serial eccentric Elon Musk reposted a video on X, the social media platform he owns. There’s nothing exciting about that, or even surprising: Musk was a habitual user even when X was called Twitter. But the video Musk shared wades into the complex debate about AI safety, deepfakes and election misinformation, because it was a digitally altered version of a video originally released by Vice President Kamala Harris’ election campaign. And Musk didn’t point this out.
In the altered video, Harris’ voice has been dubbed and her original speech has been replaced by new words that are utterly in opposition to the inspirational message of the original. There are “some people who think we should be a country of chaos, of fear, of hate. But us, we choose something different. We choose freedom,” the original begins.
Instead, in the doctored version Harris’ faked voice spews words alleging President Biden is a deep state “puppet,” is senile and that this fact was exposed during the Presidential debate. The faked version also has the Vice President saying she doesn’t “know the first thing about running the country,” the New York Times reported.
Perhaps most insultingly, the faked audio also says that Harris’ status as a person of color and a woman means she’s the “ultimate diversity hire.” Remember, this is a person who’s been Vice President for four years, has a degree in political science and economics, earned a law degree from the University of California Law School in San Francisco and won election to serve as District Attorney of San Francisco, the Attorney General of California, and as a United States Senator for California. In retweeting the video Musk noted he thought it was “amazing” and added a laughing emoji.
The nuance in this news is that when the original poster, X user “MrReaganUSA” uploaded the video it was labeled “Kamala Harris Campaign Ad PARODY.” Note the capitals. When Musk shared it, X’s system showed it had been reshared from another X user, but under his tweet the video can be viewed without seeing the uploader’s original post or text. And Musk’s tweet doesn’t mention either the parody status, nor any acknowledgement that it’s a piece of deepfake content.
It’s easy to imagine Musk, a divisive contrarian, throwing up his hands and decrying a world where he’s expected to acknowledge something that to him is a “joke” is actually fake video. But, as Engadget points out, X’s own terms and services include rules that users “may not share synthetic, manipulated, or out-of-context media that may deceive or confuse people and lead to harm (‘misleading media’).”
Some misleading deepfakes are allowed to stay live on X, Engadget says, without being branded as such since they’re memes and satires–but only if they don’t “cause significant confusion about the authenticity of the media.” In April Musk named a new head of safety after the post had been vacant for nine months–the former head of the team, then named “trust and safety” left X in June 2023 mere days after Musk complained about how the platform handled trans topics.
It’s hard not to see that in a critical election year, a faked video that corrupts not just any old content, but an official media release from a candidate, could easily cause reputational harm–not least because the clip’s now been viewed over a hundred million times, and Musk himself has incredible influence, with nearly 200 million followers on X.
At the very least, Musk is playing with fire here. There is a tightening legal focus on the matter of damaging deepfake content, which is largely made possible because of the advances in AI-powered image, video and audio manipulation. AI experts have themselves called for regulation, X rival Meta now requires posts with deepfaked content to carry a “made with AI” label. After a scandal erupted over a robocall with deepfake audio purporting to be President Biden involving voters ahead of the New Hampshire primaries, authorities and the White House have moved to fight maliciously manipulated content.
Some may dismiss Musk’s tweet as the result of his puerile sense of humor, but as owner of an influential social media platform, Musk has some responsibilities to shoulder. And as a business leader, it’s an excellent example of how not to behave if you want to inspire your staff to deliver their best work.
Refreshed leadership advice from CEO Stephanie Mehta