How AI could steal your password by listening to the sound of typing over Zoom as researchers find it can be 95 per cent accurate
- Academics have shown microphones can detect typing patterns on computers
- It could leave people open to attacks and vulnerable to having passwords stolen
Artificial Intelligence could now be advanced enough to steal passwords by listening to the sound of typing over Zoom and other recording devices, research has found.
Academics from Durham, Surrey and Royal Holloway universities have shown that microphones can detect typing patterns, which suggests anyone using a laptop in public could have their typing recorded and decoded.
Researchers trained an AI model by pressing the 36 keys on a MacBook Pro 25 times and recording the sound, before feeding it into the AI which was able to correctly identify the pattern of each key.
To prove their theory, they placed an iPhone 17cm away from the same MacBook to record someone typing, and were able to deduce the content of the typing with 95 per cent accuracy.
They also put Zoom to the test in the same way, which recorded the key strokes with an accuracy of 93 per cent.
(Stock Photo) Researchers trained an AI model by pressing the 36 keys on a MacBook Pro 25 times and recording the sound, before feeding it into the AI which was able to correctly identify the pattern of each key
(Stock Photo) Zoom was put to the test in the same way, which recorded the key strokes with an accuracy of 93 per cent
Dr Ehsan Toreini, from the Surrey University’s Centre for Cyber Security, told The Times: ‘Each key has a unique audio, or voice, that can be fingerprinted to infer what is being pressed.’
A similar study was done on an Enigma machine some years ago, which found that if it had been possible to implant a microphone, a similar outcome could have been achieved, but only with 70 per cent accuracy.
‘It gives you a hint of the tremendous improvement that has happened in the past five years in terms of the accuracy of the models, which somehow elevated the accuracy from 70-ish per cent to around perfect results,’ Dr Toreini added.
He suggested that Apple should consider adding random noises into keystrokes to ward off ‘side channel’ attacks, and that technology such as Zoom should compress audio.
Side channel attacks attempt to get signals from a device and use acoustics, power consumption and electromagnetic waves.
