Visceral Notices and Privacy Mechanisms for Eye Tracking in Augmented Reality.

Area of Science:

• Human-Computer Interaction
• Augmented Reality
• Data Privacy

Background:

• Head-worn augmented reality (AR) systems are advancing with improved power efficiency, AI, and user interaction.
• Eye-tracking sensors are crucial for AR advancements but generate sensitive data not fully understood by users.
• Existing research on privacy mechanisms for eye-tracking data focuses on technical attacks, not subjective user perception.

Purpose of the Study:

• To develop and evaluate visceral visualizations that enhance user awareness of eye-tracking data privacy in AR.
• To assess user perceptions of privacy noise mechanisms applied to visualized gaze data.
• To understand the subjective influence of privacy mechanisms on users' willingness to share gaze data.

Main Methods:

• Created visceral visualizations to represent eye-tracking data and increase privacy awareness.
• Applied privacy noise mechanisms (Weighted Smoothing, Gaussian Noise) to gaze data.
• Evaluated user comfort levels and data-sharing attitudes before and after experiencing visualizations and privacy mechanisms.

Main Results:

• Despite high initial privacy concerns, 47% of participants were comfortable sharing raw eye-tracking data.
• With privacy noise mechanisms, 70-76% of participants felt comfortable sharing gaze data.
• Overall data-sharing comfort decreased after experiencing visualizations and privacy mechanisms, yet participants remained willing to share.

Conclusions:

• Increased user understanding and access to privacy mechanisms are vital for gaze-based AR applications.
• Further research is needed to create visualizations that clearly communicate the sensitive inferences possible from raw gaze data.
• The study's codebase will be open-sourced to aid AR developers in informing users about privacy risks and providing access to privacy controls.