Mark Zuckerberg may be famed for looking and acting like a (infiltrate human society, but a team of artists working with advertising company Canny appears to built a software version of the Facebook CEO that could one day threaten to make hardware Zuckerberg obsolete.
The artists in question, Bill Posters and Daniel Howe, said they modified a a September 2017 statement Zuckerberg gave on Russian election interference on the Facebook platform using CannyAI’s “video dialogue replacement (VDR)” technology, according to Motherboard. The result is something that can’t yet walk like a Zuck or talk like a Zuck, but certainly appears to be a Zuck so long as you leave the sound off.
In their video below posted to Instagram, Deepfake Zuck—who looks quite a bit like a Weekend At Bernie’s-style corpse-marionette, though only sounds vaguely like a malicious being speaking through said flesh puppet’s dead vocal cords—brags to the audience of his unparalleled power, all courtesy of an evil organisation called “Spectre”:
Imagine this for a second: One man, with total control of billions of people’s stolen data, all their secrets, their lives, their futures. I owe it all to Spectre. Spectre showed me that whoever controls the data, controls the future.
(The audio is of an actor doing a Zuckerberg impression, not an algorithm that has a cold, for the record—though AI company Dessa recently released deepfaked audio that sounds eerily like podcaster Joe Rogan.)
The Zuckerberg deepfake is part of an exhibition called Spectre that ran in the UK’s Sheffield Doc Fest, according to Motherboard, and that has featured similar deepfakes of Kim Kardashian and Donald Trump bragging about their involvement with the fictional evil organisation.
Canny’s founders Omer Ben-Ami and Jonathan Heimann previously told FXGuide their work builds on prior research done by University of Washington researchers who helped director Jordan Peele create a deepfaked Barack Obama video warning of “fucked up dystopia.” They also told Motherboard they learned from Stanford’s Face2Face project.
All told, according to Motherboard, Canny says it generated the video from a single 21-second clip of Zuckerberg talking, paired with footage of an actor.
That the creators are uploading these videos to Facebook subsidiary Instagram is notable because the company recently refused to take down altered clips of Democratic Speaker of the House Nancy Pelosi designed to make her seem drunk or suffering the symptoms of dementia. (The video of Pelosi was not a deepfake, but instead a low-effort edit of genuine material.) Instead, Facebook resorted to a bunch of confusing half-measures, including inserting links to fact-checking websites and implementing ambiguous procedures meant to limit its reach.
According to the New York Times, Facebook spokeswoman Stephanie Otway said that the platform would apply similar procedures in this case.
“If third-party fact checkers mark it as false, we will filter it from Instagram’s recommendation surfaces like Explore and hashtag pages,” Otway told the paper.