"Reference News" recently published an article "Artificial intelligence can steal passwords by listening to your typing on the keyboard" on the website of the British "Times". The summary of the report is as follows:
Experts from Durham University, the University of Surrey and Royal Holloway University of London pressed each of the 36 keys on an Apple MacBook Pro laptop 25 times and recorded the sound. This information is then fed into an artificial intelligence program so that it can recognize the sound pattern of each key.
They then placed an iPhone 17 centimeters from the same MacBook in order to record the sound of someone typing. They managed to infer what was typed with 95 percent accuracy. When they recorded with the Zoom conferencing software, the accuracy dropped to 93 percent.
One of the authors of the research paper, Ihsan Toreni from the Center for Cyber Security at the University of Surrey, said: "Each key makes a unique sound, which can be recorded and used to infer which key is being used. Press."
Toreni said: "We use the most advanced model at present, which allows you to experience the huge improvement in the accuracy of artificial intelligence models in the past 5 years, which has made the accuracy increase from about 70%. to the point of near perfection.”
This means that techniques for carrying out "side-channel" attacks are now commonplace.
"Side-channel" attacks are those that attempt to steal signals from communication devices, potentially exploiting electromagnetic waves, acoustics, and power consumption. Apple may consider adding random noise to keystrokes to thwart such attacks, Torreni said.
The researchers also said the target's smartwatch could have been compromised and used to record keystrokes. Scientists have previously demonstrated that typing can be identified with 93.75 percent accuracy by analyzing wrist movements recorded by a smartwatch.
