Computer Mouse Can Become a Hacker’s Eavesdropping Tool — “Mic‑e Mouse” Technique Can Listen to Your Conversations
Researchers warn malware can exploit sensitive mouse sensors to pick up tiny vibrations from speech; attackers may reconstruct words with AI.
A newly identified surveillance technique, nicknamed “Mic‑e Mouse,” can turn an ordinary computer mouse into an eavesdropping device, researchers report. By exploiting the extreme sensitivity of modern mouse motion sensors, attackers may be able to capture tiny vibrations caused by nearby speech and reconstruct portions of conversations.
How the attack works
Security researchers say the attack typically follows these steps:
-
Malware installation — An attacker first installs special software (malware) on the victim’s computer. That program reads raw data from the mouse’s optical or motion sensor.
-
Vibration capture — When a person speaks, their voice causes minute air and surface vibrations. If the mouse is sitting on a hard, flat surface, those vibrations travel into the mouse and are picked up by its sensors.
-
Data filtering and cleaning — The collected vibration data is passed through signal‑processing filters to remove noise and interference.
-
AI reconstruction — Cleaned vibration traces are fed into an artificial‑intelligence model which attempts to convert the vibration patterns back into words or phrases.
Effectiveness and risks
According to the published research summary cited in media reports, the technique can recover spoken content with notable success in favourable conditions — researchers reported approximately 61% accuracy for reconstructed speech in their experiments. Numeric information such as credit card numbers or PINs may be especially vulnerable because digits produce distinct vibration patterns.
Conditions that affect success
• Surface: The mouse must be on a hard, flat surface. Soft surfaces (fabric, some mouse pads) greatly reduce vibration transmission and make the attack unreliable.
• Background noise: High ambient noise or competing vibrations (e.g., loud TV, fans) make it harder for the AI to separate speech patterns.
• Distance and volume: Loud or nearby speech produces stronger vibrations and is easier to recover.
Precautions and mitigation
Security experts recommend simple steps to reduce risk:
• Use a soft mouse pad or place the mouse on a non‑rigid surface when possible.
• Keep systems and drivers up to date and install software only from trusted sources.
• Limit unnecessary permissions and monitor processes that access HID (Human Interface Device) data.
• Treat unknown USB devices or peripherals with caution. For sensitive conversations, consider physical device control (remove or unplug peripherals) and endpoint monitoring.
While full, reliable voice reconstruction in all real‑world environments remains challenging, the discovery highlights a novel class of side‑channel threats: everyday peripherals can leak private information in unexpected ways. Further research and practical guidance from vendors and security teams are needed to assess how widespread the risk is and to develop robust defenses.
