Facebook will ban ‘deepfake’ content
AI-manipulated videos will still be allowed for parody, satire, firm says
WASHINGTON—Facebook said Monday that it would ban videos that are heavily manipulated by artificial intelligence, known as deepfakes, from its platform.
In a blog post, a company executive said Monday evening that the social network would remove videos altered by artificial intelligence in ways that “would likely mislead someone into thinking that a subject of the video said words that they did not actually say.”
The policy will not extend to parody or satire, said the executive, Monika Bickert, nor will it apply to videos edited to omit or change the order of words.
Bickert said all videos posted would still be subject to Facebook’s system for fact-checking potentially deceptive content. Content that is found to be factually incorrect appear less prominently on the site’s newsfeed and is labelled false. The new policy was first reported by the Washington Post.
Facebook was heavily criticized last year for refusing to take down an altered video of U.S. Speaker Nancy Pelosi that had been edited to make it appear as though she was slurring her words.
At the time, the company defended its decision, saying it had subjected the video to its fact-checking process and had reduced its reach on the social network.
It did not appear that the new policy would have changed the company’s handling of the video with Pelosi.
The announcement comes ahead of a hearing before the U.S. House energy and commerce committee on Wednesday morning, during which Bickert, Facebook’s vice-president of global policy management, is expected to testify on “manipulation and deception in the digital age,” alongside other experts.
Because Facebook is still the No. 1 platform for sharing false political stories, according to disinformation researchers, the urgency to spot and halt novel forms of digital manipulation before they spread is paramount.
Computer scientists have long warned that new techniques used by machines to generate images and sounds that are indistinguishable from the real thing can vastly increase the volume of false and misleading information online. And false political information is circulating rapidly online ahead of the 2020 presidential elections in the United States.
In late December, Facebook announced it had removed hundreds of accounts, including pages, groups and Instagram feeds, meant to fool users in the United States and Vietnam with fake profile photos generated with the help of artificial intelligence.