Researchers at the University of Massachusetts Lowell have demonstrated an interesting (and slightly scary) technique for using Google Glass to detect phone PINs with 83 percent accuracy from across a room – even when the screen wasn’t visible.
The technique used applies an image-recognition algorithm that doesn’t need direct sight of the screen. Instead, it uses a reference image of the target device to detect the angle at which it’s being held, then tracks the shadows from finger taps to detect which on-screen keys are being pressed …
“I think of this as a kind of alert about Google Glass, smartwatches, all these devices,” says Xinwen Fu, a computer science professor at UMass Lowell who plans to present the findings with his students at the Black Hat security conference in August. “If someone can take a video of you typing on the screen, you lose everything.”
The team also tested an iPhone 5 camera and Logitech webcam as the video feed, getting even higher accuracy. A $700 Panasonic camcorder with optical zoom was even able to grab the pin from a distance of 44 metres (144 feet) – that’s from two floors up across a street! Glass, however, would allow the most discreet form of surveillance.
The academics haven’t yet tested the approach on more complex passcodes, but estimate that an eight-character alphanumeric passcode could be detected with 78 percent accuracy.
With fingerprint login growing in popularity, the issue may not be around for long. But in the meantime, if you have sensitive data on your device, exercise care with your logins …