A recent study by UK universities demonstrated the feasibility of creating an artificial intelligence (AI) capable of deducing passwords by analyzing the sound of keystrokes as they are typed. Researchers revealed that this technique achieves a success rate of up to 95%, especially when training the program using a cellphone to capture the sounds.

However, this innovation raises concerns about cybersecurity as the increased use of tools like Zoom and the proliferation of devices with embedded microphones amplify the risk of sound-based cyberattacks.

You can also like: [VIDEO] He killed his ex-partner's fiancé, then comsuicidemitted

In an experiment conducted by experts from the universities of Surrey, Durham, and Royal Holloway in London, all keys of a MacBook Pro were pressed multiple times, varying fingers and pressures. The resulting sounds were captured through Zoom calls and with a smartphone placed near the keyboard.

By developing a machine learning system, the program was taught to recognize the acoustic characteristics associated with each key. This technique demonstrated accuracy rates of 95% in phone recordings and 93% in recordings through Zoom.