AI Identifies Passwords from Sound of Keys Typed

Must know quick bytes

• AI can identify passwords from sound of typing

• Open AI GPTBot Web Crawler Announced

• Nvidia Unveils AI Superchip

AI can identify passwords based on the sound of the keys being pressed. Scientists developed an AI system that can decipher keystrokes and identify passwords with over 90% accuracy using only audio recordings of typing. It works by listening to subtle sounds like keys clacking.

The spread of video chat apps, such as Zoom and microphone-bearing gadgets has made AI-powered sound-based hacking more feasible. This research proves even complex passwords are vulnerable to AI discerning keystrokes through audio signals.

Researchers collected audio data of typing each MacBook Pro key 25 times on Zoom and near a smartphone. This data trained a machine learning model to recognize unique acoustic features of individual keys, likely using clues like a key's position.

Tested on new audio, the AI system correctly identified struck keys 95% of the time from smartphone recordings and 93% from Zoom.

The research shows AI can infer passwords by distinguishing subtle audio differences between keystrokes, even in muffled Zoom call recordings. By learning keyboard acoustic patterns, the model accurately pinpointed pressed keys from audio alone.

You can read the full article here.