Industry News

A researcher at Carnegie Mellon University in Pittsburgh has found improved facial-recognition software could pose a significant internet security risk.

Prof. Alessandro Acquisiti's study shows facial-recognition technology, used in conjunction with people's Facebook profiles, can identify individuals and extrapolate the first five digits of their Social Security numbers.

Acquisiti uploaded webcam photographs of 93 Carnegie Mellon students to a cloud computer that compared the snapshots with over 250,000 photos in a database of publicly available images taken from Facebook profiles. With a success rate greater than 30 percent, the computer returned results matching the webcam photo with the corresponding student's Facebook picture.

Acquisiti's paper also shows that after identifying people on Facebook, he was able to correctly identify the first five digits of their Social Security numbers 27 percent of the time using information gleaned from their profiles, according to The Wall Street Journal. The Journal reports that in June, the Social Security Administration adjusted and randomized its numbering system to make numbers more difficult to steal.

Facebook uses facial-recognition technology to suggest users "tag" their friends in photos. After criticism from Connecticut Attorney General George Jepsen last June, the social media site ran ads highlighting how users can opt-out of the tag suggestion feature.