Researchers Demonstrate New Attack That Can Deduce Keystrokes from Typing Sounds
-
Researchers demonstrated a new acoustic side-channel attack that can deduce keystrokes from typing patterns, even in noisy environments.
-
The attack uses recordings of typing sounds and typing patterns to build a statistical model that predicts a target's keystrokes.
-
It achieved an average 43% typing prediction accuracy under poor conditions, not requiring controlled settings or specific keyboards.
-
The attack trains on a target's typing samples to account for normal variations, allowing some deviation in timing between keystrokes.
-
Prediction accuracy ranged widely across test subjects, from 15% to 85%, making some people more susceptible targets than others.