“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.”
Those sinister words came from the mouth of Facebook CEO Mark Zuckerberg… or so you’d think. The video was actually a deepfake made for a much different purpose. Take a look.
How will Facebook respond?
The deepfake video was made to challenge Facebook’s policy for removing fake content. Recently, the social media giant was in hot water for not removing a manipulated video of Nancy Pelosi.
Facebook has been trying to do more to stop the spread of fake news. They have increased their reporting tools for users so that they can work quickly to get the content flagged.
That Pelosi video had millions of views. Although Facebook eventually labeled the video as “fake,” they did not remove it from the site. However, the video got thousands of views before getting flagged. Now, Facebook is left to figure out if they will remove the video, and how long will they wait before they take action.
Many individuals, including Pelosi herself, called out Facebook for not removing that video. However, Facebook stuck by its guns.
“If it was the same video, inserting Mr. Zuckerberg for Speaker Pelosi, it would get the same treatment,” said Facebook director of public policy Neil Potts at a parliamentary hearing.
With this new deepfake video of Zuckerberg, we will see if what Potts said was true. Will Facebook let the video stay up? Or will it take a different stance when the CEO is the target?
How to spot deepfake videos
With the next presidential election on its way, deepfake videos could potentially sway thousands of voters. Luckily, researchers from Dartmouth College are creating software that can help spot deepfake videos.
Essentially, the software analyzes hours of authentic footage of political influencers like Pelosi, Donald Trump, and Joe Biden. The software pays attention to things like how their head moves, speech patterns, and facial expressions so that it can tell if something is authentic.
According to the researchers, the software is about 95% accurate, but they are optimistic that they can get it up to 99% within the next six months. Until detection software can outpace the fake video producers, a quick trick is to look at the eyes. Deepfake-created humans almost never blink.