The GAZEploit attack reveals serious vulnerabilities in the Vision Pro headset, allowing high accuracy in predicting what a user is typing based solely on gaze patterns.
Zhan states that distinct gaze patterns emerge when users type on the Vision Pro, with eye-fixation behavior becoming highly regular, allowing for accurate predictions of keystrokes.
Wang elaborates that these eye movements exhibit more consistent behavior while typing compared to other tasks like web browsing, highlighting a unique focus during typing.
The research achieved impressive accuracy: 92.1% for messages and 77% for passwords, indicating a critical flaw in the typing security of the Vision Pro.
Collection
[
|
...
]