I’m working on a clip-on wearable AI that uses computer vision to generate real-time “social + environment” signals (attention/glances, basic emotion cues, gestures, plus things like noise/air quality depending on the mode).
The part I’m most focused on is privacy architecture: the device processes frames locally and discards them instantly. No photo library, no video archive, no “upload later.” It’s meant to behave more like a sensor than a camera.
Questions for people who care about privacy and security: What would you personally need to see to believe “no frames are stored” is true?
[link] [comments]