Researchers at the University of Cambridge have developed an artificial intelligence tool that can recover what users type on their smartphones or tablets by listening to the interaction between their fingers and the screen. They trained a machine learning algorithm with 21 hours of audio recordings of finger taps, and the way this technology works is that it analyses sound waves that the user’s finger makes when tapping on the screen. By listening to distortions in those sound waves, the algorithm can determine the location of the finger on the screen and translate that into letters and numbers.
The tool showed moderate success during testing on a smartphone by correctly retrieving four-digit PIN codes 61 per cent of the time within 20 attempts. Also, the algorithm correctly guessed nine codes that were between seven and 13 characters long in 50 attempts carried out on a tablet.
Eavesdropping isn’t a threat just yet
For now, users shouldn’t be worried that hackers will eavesdrop on their smartphone finger taps, as that would be a harder method to deploy compared to other hacking techniques. But the researchers did demonstrate yet another way in which our devices could be compromised.