Deepfake video of Mark Zuckerberg presents a new challenge for Facebook



[ad_1]

The video, posted on Facebook this week by an Instagram account, falsely shows that Zuckerberg would have said: "Imagine this for a second: a man, with total control of billions of stolen data, their secrets, their lives, their to come up."

The video was made taking footage of Zuckerberg in 2017 and using artificial intelligence technology, known as deepfake technology, to manipulate Zuckerberg's face and make him believe that he had said something that he had not said. Zuckerberg's voice is replaced by that of an actor.

The video was created by an Israeli start-up called Canny AI, co-founder Omer Ben-Ami told CNN Business on Tuesday. The video and several others featuring celebrities such as Kim Kardashian and US President Donald Trump are part of a commissioned art installation called Spectrum, presented at the Sheffield Doc / Fest from June 6 to 11. The videos, which Canny and a few other companies have worked with artists, aim to show how technology can be used to manipulate data.
Last month, Facebook refused to remove a manipulated video that suggested that House Speaker Nancy Pelosi has escaped her words. The company said it had downgraded the fake video, which means it would be visible in the Facebook news feed of fewer people. Pelosi then criticized the company. "I think that they have proven – by not noting anything wrong that they know, that they were willing to facilitate the interference of Russia in our election", did she told KQED in California.

Asked if a falsified video of Zuckerberg would have the same treatment as Pelosi's manipulated video, Neil Potts, public policy director for Facebook, said at a parliamentary hearing held in Canada last month, "If that's the case," he said. was the same video, inserting Mr. Zuckerberg as President Pelosi, he would receive the same treatment. "

Pelosi's video was not a profound imitation like Zuckerberg's new music video, but it was manipulated using traditional video editing techniques. CNN Business asked Facebook if it had specific rules for deepfakes.

Zuckerberg's video, which was first reported by Vice, comes as the US Congress prepares to hold its first hearing on the potential threats posed by deepfake videos. Earlier this year, the US director of national intelligence warned that US opponents could use deepfake technology in future misinformation campaigns targeting the country.
Nancy Pelosi states that

Until recently, video hoaxes were relatively rare, as they are more difficult to remove than fake still images, but this is changing rapidly thanks to the rise of generative retrograde networks, or GANs. GANs can use data to produce new things. The technique is also used to make deepfakes.

In this case, said Ben-Ami, Canny AI chose a scene of less than a minute from Zuckerberg and used a computer to merge it with an actor's voice and appropriate facial movements. It took about a day to make a first version of the video, one of the artist's voices replacing Zuckerberg's impression and the facial movements of a voice actor, did he declare. It took another two or three hours to make the final version.

CNN Business contacted the artists behind Specter but did not immediately receive an answer.

Ben-Ami, whose company focuses on dubbing speeches in videos from one language to another, is concerned about video and by others that Canny AI has aired without the context in which they were created. But he added that he also wanted to ask about the creation of such media.

"People need to know that it is possible to do it," he said.

An Instagram spokesperson told CNN Business that the site will process the video "in the same way that we handle all the wrong information on Instagram". If it's marked as fake by third-party fact checkers, the spokesperson said, the site's algorithms will not recommend posting by the netizens.

Last month, Facebook announced that it would spend millions of dollars to fund academic institutions studying the analysis of images and videos. Part of this work involves attacking deepfakes, the company said.

[ad_2]

Source link