We all know Facebook is an easy privacy punching bag. The social network has a tendency to tinker with our personal data, and deservedly lands in hot water now and then with paranoid users, pundits and politicians.
But the latest episode in Facebook’s ongoing privacy drama has little to do with Facebook itself. Researchers at Carnegie Mellon University’s Heinz College have discovered how to use Facebook profile pictures and facial recognition software to match people’s names to their faces, with a success rate of 31%, according to CNet.
Here’s how it worked: The researchers created a database of 25,000 Facebook profile pictures from students at the university. Then, they gathered up some volunteers and snapped some photos with a webcam. After about three seconds of rapid-fire comparisons, the software was able to identify the student nearly a third of the time.
In another experiment, the researchers compared 277,978 Facebook profile photos to nearly 6,000 profiles from a dating site. Through facial recognition, the researchers identified roughly 10% of the dating site’s users — many of whom used pseudonyms.
The implications are kind of frightening. “Google’s Eric Schmidt observed that, in the future, young individuals may be entitled to change their names to disown youthful improprieties,” the researchers noted. “It is much harder, however, to change someone’s face.” The study envisions a future where one piece of data—even a face—can lead to a trove of personal information.
Still, this isn’t Facebook’s problem, at least not solely. Although Facebook employs facial recognition to automatically tag users in their friends’ photos, the researchers used their own methods to gather up photos and perform face matching on their own. In this case, Facebook’s biggest fault is that it makes basic profile information—including a name and a profile picture—available to anyone through a simple web search. But that’s also true of Twitter and of Google, which recently decided that all Google Profiles must be public.
So although Facebook formed the basis of Carnegie Mellon’s research, the study is more of a general cautionary tale about facial recognition and the amount of information we make publicly available on the Internet. Over time, face-matching will only become more accurate, leading to better results than the 31% in this study, but that doesn’t mean Facebook is doing something wrong by allowing facial recognition in friends’ photos.
If, however, Facebook comes out with a tool to identify anyone in its network just by snapping a picture of some dude on the street, then users should take up their privacy pitchforks once again.