Identity-Decoupled Anonymization for Visual Evidence in Multi-modal Retrieval-Augmented Generation

2026-04-26

AI Summary

This work addresses the challenge of preserving privacy in multimodal retrieval-augmented generation systems, where retrieved images often contain identifiable facial information. Existing anonymization techniques either degrade non-identity visual semantics essential for downstream tasks or fail to provide rigorous privacy guarantees. To overcome this, the authors propose inserting a generative anonymization module between retrieval and generation stages. This module employs an identity-attribute disentangled variational encoder, a manifold-aware identity replacement sampler, and a conditional diffusion generator distilled into a latent consistency model, collectively replacing identity features while preserving structured attributes. By integrating an ensemble of identity recognizers with a hinge loss, the method achieves provable privacy protection. The approach effectively eliminates identity identifiability without compromising visual fidelity, thereby maintaining strong downstream task performance and supporting low-latency deployment.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Privacy
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Security and Privacy: Data transparency and provenance

Abstract

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.

Problem

• multi-modal retrieval-augmented generation
• visual evidence
• face anonymization
• privacy preservation
• identity information

Innovation

• disentangled representation
• generative anonymization
• retrieval-augmented generation
• latent diffusion model
• privacy-preserving AI