Explore how diffusion models generate realistic synthetic gaze data while attenuating privacy-sensitive internal states, enabling secure and scalable eye-tracking applications.