In an early morning message on January 8, Facebook said it would ban the release of videos that were deeply manipulated by artificial intelligence. It is the latest in a series of tweaks the company has made to crack down on fake information on the platform. On Monday, Monika Bickert, a Facebook executive, said in a blog post that facebook would remove videos commonly referred to as “deep fakes.”
In such videos, a changes made by artificial intelligence “are likely to mislead people into thinking that a subject in the video says something that doesn’t actually exist.” Facebook ads will also ban such videos.
However, industry insiders say the policy has limited effect in controlling the spread of fake videos, the vast majority of which are edited in the traditional way: manually deleting background information or adjusting the order of language. Mr Bickert said Facebook’s new policy would not cover such videos and would not affect spoofing or satirical videos.
She said all uploaded videos would still be checked by Facebook’s fact-checking system to avoid possible deceptive content. If the content is found to be inaccurate, it will not be highlighted in the message flow, but also will be labeled.
Facebook, the world’s largest social network, is trying to crack down on fake news spreaders ahead of this year’s US presidential election, according to the latest move. In the 2016 U.S. election, fake news went viral on Facebook, leading to widespread criticism of the company.
Facebook’s new policy is also aimed at appeasing lawmakers, academics and political campaigns. They remain disappointed with the way Facebook handles political content.
Some Democratic politicians argue that Facebook’s new policy is not strong enough. Last year, a controversial video by House Speaker Nancy Pelosi, but Facebook refused to remove it. The company defended its decision, saying it had fact-checked the content of the video and narrowed its reach on social networks.
The new policy does not apply to such videos. Fake information researchers refer to such videos as “light-based phishing” videos, which use simple video editing software to edit deceptive content, which is still quite different from deep-sourced phishing videos produced by artificial intelligence.
Drew Hammill of Pelosi’s team said in a statement that Facebook “wants you to think that the problem is in video editing, but the real problem is Facebook’s refusal to stop the spread of fake information.”