Last summer a video of Mark Zuckerberg went viral. It seemed to show the Facebook founder talking like a megalomaniac. Two artists made it, called it satire, and Facebook left it online at the time.
Whether a clip like that would survive now is an open question. Facebook announced a policy update late on 6 January 2020, and Monika Bickert, the company’s VP of global policy management, wrote that it will take a stricter line on manipulated media. Out go videos that have been edited or synthesized “in ways that aren’t apparent to an average person and would likely mislead someone into thinking that a subject of the video said words that they did not actually say.”
What the ban covers
The rule also reaches content made with artificial intelligence that will “merge, replace or superimpose content onto a video, making it appear to be authentic.” TechCrunch noted that this probably catches the crudest material, such as revenge deepfake porn, where a real person’s face is pasted onto an adult performer’s body. Nudity was already banned on the platform.
Photorealistic faces generated by AI, people who look real but are not, also fall under the heightened scrutiny.
What it leaves alone
The ban is not total. Parody and satire are carved out. So are edits made for quality, and cuts or splices that shorten a clip or reorder the words in it.
TechCrunch pointed to what that omission means in practice. During the recent UK general election, campaign staff for one party edited a video of a rival politician who had been asked a question about Brexit, making it look as though he had no answer when he did. Doctoring of that kind would go untouched under the new policy. The outlet read the move as Facebook chasing quick wins on the most obviously fake material, where pulling a post is unlikely to draw accusations of partisan bias.
The same outlet expected deepfake makers to keep probing the edges of the rule to see what they can get away with.
