Facial recognition software could reveal your social security number

The study, led by Alessandro Acquisti from Carnegie Mellon University, combined the use of three different technologies - cloud computing, facial recognition and public information that can be found on various social networking sites.

They used these technologies in three different experiments. In the first experiment, Acquisti and his team were able to identify members of an online dating site where members do not use their real names for identification. The second experiment allowed the research team to identify college students in real life walking on campus based solely on their face and information gathered online.

In the third experiment, the researchers used the technology to predict personal interests and identify students, including some social security numbers, with only a photo of their face to start. Social security numbers, Acquisti pointed out in 2009, are a security flaw as they can be predicted if you know the person’s hometown and date of birth. This new technology uses that information and in many cases can determine a person’s social security number. In this experiment, the researchers looked at Carnegie Mellon University students and those who had a date of birth and hometown displayed on their social media account page.

They developed a smartphone application which gathers both online and offline information and displays it over the person’s facial image on the phone.

Acquisti says that while all attention has been turned to providing security in cyberspace, this new technology allows one to step outside of cyberspace and into the real world. Using cyberspace as a means to identify someone living down the street or who you pass by every day on your way to work is a possibility. The idea of protecting your privacy has been changed and caution should be used before you share the next picture of yourself.

Facebook has recently limited the facial recognition photo-tagging on its site to people in your friends list and Google has not yet allowed facial recognition technology into its searches.

More information: "Identifying indicators of illegal behaviour: carnivore killing in human-managed landscapes," Proc. R. Soc. B July 27, 2011. doi: 10.1098/rspb.2011.1228

© 2010 PhysOrg.com