Researchers at the University of California, Irvine, have discovered a new method that bad actors could use to collect data from unsuspecting persons. Dubbed the Mic-E-Mouse method, the attack involves hacking into a user’s high-performance optical mouse (with 20,000 DPI or more) and using the mouse’s sensors to pick up vibrations in the air caused by speech. These signals are then run through a program that leverages machine learning to translate the vibrations to speech, ultimately allowing eavesdropping through a computer mouse.
“The main two parameters we look at in the mouse are the sampling rate and the DPI,” researcher and PhD student Mohamad Fakih told The Register. “And it’s capable of picking up more than just speech. If there’s anyone in the room and their steps are causing vibrations, you can track their movement.”
