MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents

2026-05-10

AI Summary

This work addresses the privacy risks inherent in personalized memory within edge–cloud intelligent systems, where existing approaches often compromise semantic fidelity through excessive masking. To resolve this trade-off, the authors propose a type-aware placeholder mechanism that identifies privacy-sensitive segments at the edge and replaces them with structured placeholders; these are processed in the cloud while enabling on-demand local restoration of original content, thereby decoupling privacy preservation from semantic integrity. The study introduces MemPrivacy-Bench, the first benchmark for evaluating privacy in personalized memory, featuring a four-tier privacy taxonomy and configurable policies. Experimental results demonstrate that the proposed method significantly outperforms GPT-5.2 and Gemini-3.1-Pro in privacy protection efficacy, achieves lower inference latency, and limits memory utility degradation to under 1.6%, markedly surpassing conventional masking strategies.

Abstract

As LLM-powered agents are increasingly deployed in edge-cloud environments, personalized memory has become a key enabler of long-term adaptation and user-centric interaction. However, cloud-assisted memory management exposes sensitive user information, while existing privacy protection methods typically rely on aggressive masking that removes task-relevant semantics and consequently degrades memory utility and personalization quality. To address this challenge, We propose MemPrivacy, which identifies privacy-sensitive spans on edge devices, replaces them with semantically structured type-aware placeholders for cloud-side memory processing, and restores the original values locally when needed. By decoupling privacy protection from semantic destruction, MemPrivacy minimizes sensitive data exposure while retaining the information required for effective memory formation and retrieval. We also construct MemPrivacy-Bench for systematic evaluation, a dataset covering 200 users and over 52k privacy instances, and introduce a four-level privacy taxonomy for configurable protection policies. Experiments show that MemPrivacy achieves strong performance in privacy information extraction, substantially surpassing strong general-purpose models such as GPT-5.2 and Gemini-3.1-Pro, while also reducing inference latency. Across multiple widely used memory systems, MemPrivacy limits utility loss to within 1.6%, outperforming baseline masking strategies. Overall, MemPrivacy offers an effective balance between privacy protection and personalized memory utility for edge-cloud agents, enabling secure, practical, and user-transparent deployment.

Problem

• privacy-preserving
• personalized memory
• edge-cloud agents
• memory utility
• sensitive data exposure

Innovation

• privacy-preserving memory
• edge-cloud agents
• semantic placeholders
• personalized LLM agents
• privacy taxonomy