Tue, 21-Oct-2025

Google Ads | Google Ads | Google Ads | Google Ads | Google Ads | Google Ads | Google Ads | Google Ads

AI Can identify Keyboard Sounds to Steal Passwords with 95% accuracy

AI can steal your Password

AI Can identify Keyboard Sounds to Steal Passwords with 95% accuracy

Initially, our concerns centered around the security of our webcams and microphones, but now even our trusty keyboards have fallen under the shadow of vulnerability. Those who rely on laptops for their daily tasks are now confronted with a new threat: the potential exposure of their sensitive information, encompassing private conversations, passwords, and credit card particulars.

A recent scholarly work, a collaborative effort by a group of esteemed British university researchers, accentuates the disconcerting reality that artificial intelligence holds the capacity to decipher our keystrokes based solely on the auditory cues they produce. This remarkable ability reaches an impressive accuracy rate of 95%. As we contemplate this revelation, it’s evident that these incursions are only poised to evolve into more sophisticated forms, fueled by the rapid strides of AI advancement.

How it Works

The research paper explores a fascinating area known as “acoustic side channel attacks.” In simple terms, this involves a sneaky technique where a malicious individual uses another device, like a cell phone placed near a laptop, or a microphone that’s not muted during a video call (like on Zoom), to capture the sound of typing.

Afterward, they take this recorded sound and run it through a sophisticated AI model that has been trained to recognize the unique sound patterns of each keystroke. This enables them to figure out exactly what was typed by listening to the sound of typing.

In their experiments, the researchers impressively managed to figure out what was typed on a MacBook Pro by just analyzing the typing sounds picked up by a nearby mobile phone. They achieved a remarkable accuracy rate of 95%. Similarly, even when analyzing typing sounds from a recorded Zoom call, they still achieved a concerning accuracy rate of 93%.

The paper says:

With the recent developments in both the performance of (and access to) both microphones and DL models, the feasibility of an acoustic attack on keyboards begins to look likely.

How to Avoid Such Attacks

A method to prevent these attacks would involve creating more robust passwords that encompass both upper and lower case letters. This way, the AI would encounter difficulty in comprehending the amalgamation of keystrokes. Furthermore, passwords composed of complete words would be simpler for unauthorized access to decipher, as opposed to a random assortment of numbers, letters, and additional symbols.

[embedpost slug=”/nokia-reveals-stylish-150-130-music-feature-phones-with-essential-features/”]