The Guardian is reporting that Facebook has banned “deepfake” videos in preparation for the U.S. election.
Deepfake videos use AI to manipulate a video to make it appear something is happening that’s really not. For example, a deepfake video may superimpose one person’s face onto another’s body, or make it appear someone is saying something they’re not. During an election, a deepfake video at just the right moment could have a profound impact on the results, swaying public opinion enough to change the outcome.
Even aside from elections, deepfake videos could have disastrous consequences. A video showing someone seemingly engaged in immoral or illegal behavior, engaging in controversial or even racist speech, could ruin a person’s life.
As a result of these threats, Facebook has decided to ban deepfake videos—with one big caveat. The company is only banning deepfake videos created with the help of AI. Videos created using traditional methods, so-called “shallow fakes,” have not been banned, despite having the ability to be almost as convincing.
While this is certainly a step in the right direction, it remains to be seen if Facebook’s efforts will be enough to have a measurable impact.