Facebook will take down videos that have been doctored so that a person appears to say words they never said. Monika Bickert, the company’s vice president of global policy management, announced the change in a policy update.
The ban reaches content “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.” It also covers material made with AI tools that “merge, replace or superimpose content onto a video, making it appear to be authentic.” Fake photorealistic faces generated by AI, and deepfakes that paste a real person’s face onto another body, fall into that group.
The rule has gaps. Edits made for quality, and cuts or splices that shorten a clip or reorder words, are not covered. Parody and satire are exempt.
Critics pointed to the recent UK general election. Campaign staff for one party edited a video of a rival politician who had been asked about Brexit, so that he looked lost for words when he was not. That kind of edit would go untouched under the new policy. A widely shared deepfake of Mark Zuckerberg last summer, which seemed to show him speaking like a megalomaniac, is also an open question. Its creators, two artists, called it satire, and Facebook left it up at the time.
Media that falls outside the ban can still be checked and have its spread limited, Bickert said, especially if it carries violating information.
