Satellites are LEAKING Your Unencrypted Data

Quick Overview

Research reveals that modern optical computer mouse sensors are vulnerable to acoustic eavesdropping attacks, named Mic-E-Mouse, which leverage sound waves to reconstruct user speech and digits with surprising accuracy, especially at higher audio volumes and polling rates.

Key Points: Researchers introduced Mic-E-Mouse, the first attack that exploits high-performance optical mouse sensors to capture and reconstruct audio signals, including speech and digits. The attack achieved 61.57% accuracy for digit recognition and 47.65% for speech recognition when audio volume was 80dB and the polling rate was 8kHz. The success rate of reconstruction is highly dependent on surface material, achieving 61.57% accuracy on plastic, compared to only 23.06% on cardboard. The researchers demonstrated that modern mice with high DPI (like 20,000 DPI) and high polling rates (like 8kHz) are significantly more vulnerable than older models. The Kido nursery data breach, where hackers extorted the company, is used as an example, although the hackers later deleted the data after public exposure. The process involves capturing raw audio via the mouse sensor, applying Wiener Filtering, and then Neural Filtering to recover intelligible speech. The video concludes by showing that data brokers collect vast amounts of personal information legally, emphasizing the need for end-to-end encryption and services like DeleteMe to combat data exposure.

Context: This video investigates security vulnerabilities in consumer electronics, specifically focusing on the often-overlooked potential for optical mouse sensors to be repurposed as covert microphones. The research, detailed in a paper titled "Invisible Ears at Your Fingertips: Acoustic Eavesdropping via Mouse Sensors," explores how sound waves vibrating a desk surface can be translated by the mouse sensor into digital data, which can then be processed to recover speech or typed digits.

Detailed Analysis

The central theme of the video is the vulnerability of modern optical mouse sensors to acoustic side-channel attacks, allowing attackers to eavesdrop on nearby conversations or keystrokes. The researchers developed an attack named Mic-E-Mouse, which leverages the high sensitivity of modern sensors (like those with 20,000 DPI and 8kHz polling rates) to capture sound vibrations transmitted through a desk surface. The attack works by capturing raw audio data via the mouse sensor, which initially appears unintelligible, then applying Wiener Filtering to remove some noise, and finally using a Neural Network to reconstruct clear speech or digits. Performance tables show that accuracy is highest (over 61% for digits and 47% for speech at 80dB) on plastic surfaces and decreases significantly on softer materials like cardboard (down to 10.69% for speech). The video also connects this technical threat to real-world data exposure by discussing the Kido nursery data breach, where hackers extorted the company using stolen children's data, and by showing how data brokers legally aggregate personal information, concluding with a promotion for a data removal service, DeleteMe.