Facebook just banned deepfakes, but the policy has loopholes — and a widely circulated deepfake of Mark Zuckerberg is allowed to stay up

source
Drew Angerer/Getty Images
  • Facebook will now remove deepfakes, the company announced Monday night. Deepfakes is a term for video or imagery manipulated with artificial intelligence to show something fake.
  • But Facebook’s new policy has considerable loopholes: It allows deepfakes meant as satire as well as misleading videos made with less sophisticated tools.
  • A widely circulated doctored video of Nancy Pelosi and a deepfake video of Mark Zuckerberg will both still be allowed under the new policy.
  • Critics – including social-media watchdogs and House Speaker Nancy Pelosi – slammed Facebook for the loopholes.
  • Visit Business Insider’s homepage for more stories.

Facebook is cracking down on misleading deepfake videos, but the company’s new ban on deepfakes has loopholes.

Deepfakes are now banned on Facebook and Instagram, the company announced in a blog post Monday night. Deepfakes is a term for video or imagery manipulated by artificial intelligence to show something that isn’t real. A widely shared example is the actor Nicolas Cage’s face being swapped onto actress Amy Adam’s body.

But Facebook’s policy doesn’t extend to deepfakes meant as parody or satire, nor does it ban other forms of doctored videos made with less sophisticated software. It also doesn’t override Facebook’s policy against fact-checking politicians, meaning deepfakes posted by a political candidate would not be removed.

While the new policy is being heralded as a step in the right direction, misinformation watchdogs are calling on Facebook to do more.

“This is a good policy,” Sam Gregory, program director at human-rights nonprofit Witness, told Business Insider. “It’s proactive in advance of an emerging threat. It recognizes that invisible-to-the-eye AI-based manipulation poses a real challenge.”

Gregory served as an adviser to Facebook as it prepared the new policy.

“But dealing with deepfakes is a step but not enough,” he added. “They need to make sure they take clearer steps to reduce spread of known false and inflammatory videos more quickly, and they need to provide better tools to users and journalists to understand how something has been manipulated.”

Even with the new policy, loopholes remain

Because the policy protects videos intended as satire, many deepfakes will remain online.

One of those is a deepfake of Mark Zuckerberg created by activists in June to protest misinformation on Facebook. The doctored video shows Zuckerberg making a fake speech about power.

A Facebook spokesperson confirmed to Business Insider that the Zuckerberg deepfake falls within the policy’s satire category.

The new policy doesn’t cover more rudimentary forms of doctored videos either. An infamous clip of House Speaker Nancy Pelosi, edited to make it look as if she’s slurring her words, will remain on Facebook even under the new policy, the Facebook spokesperson confirmed to Business Insider. Yet Facebook’s fact-checkers have marked the video with a warning that it is false.

In a statement to Business Insider on Tuesday, a spokesperson for Pelosi criticized Facebook’s decision not to remove the video.

“Facebook wants you to think the problem is video-editing technology, but the real problem is Facebook’s refusal to stop the spread of disinformation,” the Pelosi spokesperson said.

Security experts are bracing for an influx of doctored and misleading videos on social media as the 2020 election approaches, according to Vijay Balasubramanian, CEO and cofounder of fraud prevention firm Pindrop.

“The upcoming election is the perfect catalyst for a flood of these videos,” Balasubramanian told Business Insider. “Any policy to at least say that you’re taking this seriously is a step in the right direction, but I kind of wonder why the [Facebook] policy is so narrowly scoped.”

‘A step in the right direction’

Even as privacy watchdogs are calling on Facebook to continue to tighten its policies around misleading content, some heralded the new policy as an important step in the fight against misinformation online.

Danielle Citron, a Boston University law professor and vice president of the Cyber Civil Rights Initiative, advised Facebook on the new policy, and told Business Insider that she sees it as fair, but would like to see more policies banning misleading videos in the future.

“This is indeed a step in the right direction … We need to leave breathing room for satire and parody,” Citron said. “I would have liked the policy to have including manipulated deceptive media showing someone acted in ways they didn’t. Think about deepfake sex videos.” Deepfake sex videos usually involve women’s faces being pasted onto bodies of porn actresses without their consent.

Gregory emphasized that Facebook should continue to beef up its policies around “shallowfakes,” or videos that use less sophisticated technology to mislead people.

“The global problem right now is around mis-contextualized video (claiming to be from one place or time but actually from another), and … to a lesser extent lightly edited or manipulated videos,” Gregory said. “The platforms need to address this problem – not just to protect US presidential candidates but for the vast majority of the world’s population.”

Got a tip? Contact Aaron Holmes at (706) 347-1880 or at aholmes@businessinsider.com. Open DMs on Twitter at @aaronpholmes. You can also contact Business Insider securely via SecureDrop.