I remember the first time I got creeped out by Facebook’s facial recognition capabilities like it was yesterday. I was browsing my Facebook, checking my newsfeed, writing dumb statuses, and then BAM! Facebook accidentally tagged me in a really
funny unfortunate photo of my mom.
The Deepface project was officially presented in the IEEE Computer Conference in March of last year. As as Verge writer Josh Lowensohn states, “the technology maps out 3D facial features, then makes a flat model that’s filtered by color to characterize specific facial elements.”
According to Facebook, this system allows the site to recognize user’s faces with 97.5% accuracy. To put that in perspective, that is significantly more advanced than the FBI’s facial recognition system, which only guesses faces with about 85% accuracy. The reason for this is that unlike the FBI, Facebook does not rely on poorly angled photos taken at various bureaucratic interactions or taken through security cameras, but rather on the 250 billion photos that Facebook users have posted of themselves. Most of these pictures are of excellent quality, and are in fact designed precisely for members in our social network to recognize our face immediately.
To illustrate the significance of this power more tangibly, Alessandro Acquisiti provides an experiment he did with this tool (or something close to it) with students at a university (this is taken from this Ted talk):
“We asked students who were walking by to participate in a study, and we took a shot with a webcam, and we asked them to fill out a survey on a laptop. While they were filling out the survey, we uploaded their shot to a cloud-computing cluster, and we started using a facial recognizer to match that shot to a database of some hundreds of thousands of images which we had downloaded from Facebook profiles. By the time the subject reached the last page on the survey, the page had been dynamically updated with the 10 best matching photos which the recognizer had found, and we asked the subjects to indicate whether he or she found themselves in the photo… the computer did, and in fact did so for one out of three subjects.
The power of this system is striking, and the FBI and other government agencies know it.
As of June of 2014, Facebook lost a case against the Manhattan district attorney that allows the Manhattan city government to demand “nearly complete account data… ranging from pages they had liked to photos and private messages” and also to Facebook’s facial recognition system. If scaled up, this means that national agencies and state and local governments may have the right to demand this information as well. While the case is currently in appeals, it nonetheless suggests that Facebook’s remarkable facial recognition system could be in the hands of the government in just a few years.
Even if it isn’t, this still show’s how contentious and powerful facial recognition is in the digital age. On this point, I think most people, like myself, find this to be quite creepy and chilling. While of course there are very strong political and business incentives for the government and for Facebook to use this information properly, morally, and legally (i.e for security purposes, quick-tagging, etc.), the threat that they may abuse this power against our own will is a serious concern. A Harvard Law Review article entitled “In the Face of Danger: Facial Recognition and Limits of Privacy Law” makes this point clear: “as our networked world has expanded what is possible, it has been slow to catch up and determine what is permissible.” That is to say, the abuse of facial recognition, either by the government, or by Facebook, or by some other third party, may (and maybe must?) occur before any legal changes can be made to prevent it from happening again or altogether.
If these seems as crazy to you as it does to me, then you may be interested in buying these privacy glasses that prevent Deepface from recognizing your facial features. I can’t promise you that you’ll look good, but this is definitely an interesting way to protect your privacy rights, and perhaps to prevent the emergence of Skynet and the upcoming robot apocalypse.