(Dmytro “Henry” Aleksandrov, Headline USA) Recent research discovered that artificial intelligence could be advanced enough to steal the passwords of people by listening to the sound of them typing over Zoom and other recording devices.
Academics from Durham, Surrey and Royal Holloway universities realized that microphones can detect typing patterns, which automatically means that any person who is using a laptop in public could have one’s typing recorded and decoded, the Daily Mail reported.
Researchers trained an AI model by pressing the 36 keys on a MacBook Pro 25 times while recording the sound. After that, they “fed” the sounds into the AI which was able to correctly identify the pattern of each key.
To be able to prove their theory, the researchers placed an iPhone 6.7 inches away from the same MacBook Pro to record someone typing and the AI was able to deduce the content of the typing with 95% accuracy.
The researchers also tested Zoom in the same way and the AI was able to record the keystrokes with an accuracy of 93%.
“Each key has a unique audio, or voice, that can be fingerprinted to infer what is being pressed,” Dr. Ehsan Toreini, a researcher from the Surrey University’s Centre for Cyber Security, said.
Some years ago, academics performed a similar study on an Enigma machine. The researchers of that study realized that if it had been possible to implant a microphone, a similar outcome could have been achieved, even though the accuracy would be significantly lower — only 70%.
“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 percent to around perfect results,” Toreini said.
He then added that Apple should consider adding random noises into keystrokes to ward off “side channel” attacks. Additionally, Toreini suggested that technology such as Zoom should compress audio to make it harder for AI to steal your private information.
