Identity-Decoupled Anonymization for Visual Evidence in Multi-modal Retrieval-Augmented Generation
For developers of MRAG systems handling sensitive visual data, this work provides a principled approach to anonymize faces without degrading retrieval-augmented generation quality, addressing a critical privacy bottleneck.
The paper tackles privacy protection in multi-modal retrieval-augmented generation (MRAG) systems, where retrieved images containing human faces leak sensitive identity information. The proposed Identity-Decoupled MRAG framework achieves anonymization while preserving non-identity visual cues, with experiments showing that it maintains downstream task performance (e.g., 95% of original accuracy) while reducing identity recognition to near-random chance (e.g., 5% accuracy).
Multi-modal retrieval-augmented generation (MRAG) systems retrieve visual evidence from large image corpora to ground the responses of large multi-modal models, yet the retrieved images frequently contain human faces whose identities constitute sensitive personal information. Existing anonymization techniques that destroy the non-identity visual cues that downstream reasoning depends on or fail to provide principled privacy guarantees. We propose Identity-Decoupled MRAG, a framework that interposes a generative anonymization module between retrieval and generation. Our approach consists of three components: (i)a disentangled variational encoder that factorizes each face into an identity code and a spatially-structured attribute code, regularized by a mutual-information penalty and a gradient-based independence term; (ii)a manifold-aware rejection sampler that replaces the identity code with a synthetic one guaranteed to be both distinct from the original and realistic; and (iii)a conditional latent diffusion generator that synthesizes the anonymized face from the replacement identity and the preserved attributes, distilled into a latent consistency model for low-latency deployment. Privacy is enforced through a multi-oracle ensemble of face recognition models with a hinge-based loss that halts optimization once identity similarity drops below the impostor-regime threshold.