Facebook has come under fire in recent times over various privacy-related scandals, leading to some to call upon other users to delete their Facebook accounts. However, in a pretty strange twist, it seems that researchers at Facebook have managed to develop an AI system that is capable of “hiding” people from facial recognition systems.
Basically, this seems like some kind of reverse deepfake where it distorts the person’s face ever so slightly to the point where it can confuse facial recognition systems, but at the same time maintain enough of the original so that you, as a human, will definitely know who it is that you’re seeing.
For those unfamiliar, deepfake videos are where peoples faces are superimposed onto another person’s face using AI, where it can blend in seamlessly. This in turn has created some problems where deepfake videos could be used to spread false messages. However, with Facebook’s AI, it is possible to distort the faces in a way that it may no longer be useful.
According to Facebook, “Recent world events concerning the advances in, and abuse of face recognition technology invoke the need to understand methods that successfully deal with de-identification. Our contribution is the only one suitable for video, including live video, and presents quality that far surpasses the literature methods.” Unfortunately, Facebook has told VentureBeat that there are currently no plans to implement this tool into its products and services.