by Fake social media accounts are increasingly likely to show fake faces.
Facebook’s parent company Meta says more than two-thirds of the influencer operations it found and removed this year used computer-generated profile pictures.
As the AI behind these fakes has become more widely available and better at creating realistic faces, bad actors are adapting them for their attempts to manipulate social networks.
“It seems like these threat actors are thinking, This is an ever better way to hide,” said Ben Nimmo, who leads global threat intelligence at Meta.
That’s because it’s easy to go online and download a fake face, instead of stealing a photo or an entire account.
“They probably thought…it’s a person that doesn’t exist, and so there’s no one who’s going to complain about it and people won’t be able to find it the same way,” Nimmo said.
The forgeries have been used to push Russian and Chinese propaganda and harass activists on Facebook and Twitter. An NPR investigation this year found that they are also used by LinkedIn marketing scammers.
The technology behind these faces is known as a generative adversarial network, or GAN. It’s been around since 2014, but it’s gotten a lot better in the last few years. Websites today allow anyone to generate fake faces for free or for a small fee.
A study published earlier this year found that AI-generated faces have become so convincing that people have only a 50% chance of correctly guessing whether a face is real or fake.
But computer-generated profile pictures also often have telltale signs that people can learn to recognize, like quirks in their ears and hair. oddly aligned eyesand weird outfits and backgrounds.
“The human eyeball is an amazing thing,” Nimmo said. “Once you look at 200 or 300 of these AI-generated profile pictures, your eyeballs start to pick them out.”
This made it easier for researchers from Meta and other companies to locate them on social networks.
“There is this paradoxical situation where threat actors think that by using these AI generated images, they are really smart and they are finding a way to hide. We are actually sending out another signal that says, this account looks fake and you have to watch it,” Nimmo said.
He says it’s a big part of how threat actors have evolved since 2017, when Facebook began publicly taking down networks of fake accounts that secretly attempted to influence its platform. It has since removed more than 200 such networks.
“We are seeing online operations just trying to spread to more and more social media platforms, and not just for the big ones, but for the small ones as much as possible,” Nimmo said. This includes emerging and alternative social media sites, such as Gettr, Truth Social and Gab, as well as popular petition websites.
“Threat Actors [are] just trying to diversify where they put their content. And I think it’s in hopes that something somewhere doesn’t get caught,” she said.
Meta says it works with other tech companies and governments to share information about threats, because they rarely exist on a single platform.
But the future of that work with a critical partner is now in question. Twitter is undergoing major upheaval under new owner Elon Musk. He made deep cuts to the company’s trust and safety workforce, including teams focused on non-English languages and state-backed propaganda operations. The key leaders in trust and safety, security and privacy are all gone.
“Twitter is in a transition right now, and most of the people we’ve dealt with have moved on,” said Nathaniel Gleicher, head of security policy at Meta. “As a result, we have to wait and see what they announce in these threat areas.”
Copyright 2022 NPR. To learn more, visit https://www.npr.org.
