QRNG-RAW is a camera module that removes personal data before an image is formed. Positioned between the image sensor and the ISP, an AI detection layer identifies faces, license plates, bodies, and documents directly in 12-bit RAW data, before any viewable image exists. In the default track, a quantum random number generator (NIST SP 800-90A/B compliant) irreversibly overwrites those regions through triple destruction—quantum-noise substitution, bit-depth reduction, and spatial degradation—so no original ever exists to leak, while background context remains intact for security monitoring. In a legally gated second track, regions are instead encrypted with quantum-generated AES-256 keys stored in a hardware security module and automatically destroyed on expiry; every access is recorded in a tamper-evident audit log. All processing runs on-device in real time at 1080p 30 fps. The module integrates into CCTV, service robots, elevators, hospitals, and vehicles without changes to existing video-management systems.