Facebook bans certain types of deepfake videos on network
Facebook bans deepfakes from the social network. Computer-generated videos will no longer be allowed to be distributed. However, what exactly is meant by deepfakes can be interpreted broadly.
Facebook confirms the ban in a blog post, after the Washington Post reported about it earlier. In the blog post, Facebook writes what the ban will look like and what types of videos fall under it. There are two specific requirements that qualify a video as ‘deepfake’. This happens when the video is “edited—other than adjustments for clarity or quality—in a way that isn’t immediately apparent to the average person, and may mislead a person into thinking that someone on a video is saying something that wasn’t actually said.” ‘. Another requirement is that a video is “created by artificial intelligence or machine learning that merges, replaces or adds content in a video to make it appear authentic.” Facebook says parodies and satire aren’t covered by the rules. According to the company, videos in which only the order of words has been changed are not deepfakes that should be removed.
Videos that have been edited may therefore still appear on the platform. That way, the controversial video about US Congresswoman Nancy Pelosi doesn’t fall under the rules. Last year, an edited video was widely circulated in which the politician appeared to be drunk or confused. It later turned out that it was a video in which the sound had been deliberately slowed down. Facebook was criticized for letting the video stand. Facebook has also been under fire for some time for not having a conscious policy for deepfake videos, while they may be used in the run-up to the US presidential election later this year. Also, last week, a video circulated in which a speech by presidential candidate Joe Biden had been edited to make him appear to say racist things. Such videos are also not covered by Facebook’s new policy.
Facebook says it wants to collaborate a lot with experts, among others, for the new policy. The company is said to have spoken with more than 50 experts in the fields of policy, media and the law. The company has also started a collaboration with the Reuters news agency, whereby journalists worldwide can follow training to recognize deepfakes and manipulated media. According to Facebook, this collaboration is separate from current policy.