MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents
Summary
MemPrivacy is a research paper introducing a framework for privacy-preserving personalized memory management in edge-cloud AI agents, using type-aware placeholders to protect sensitive data while maintaining semantic utility. It includes a new benchmark dataset and demonstrates superior performance over general-purpose models like GPT-5.2 and Gemini-3.1-Pro.
View Cached Full Text
Cached at: 05/13/26, 04:12 AM
Paper page - MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents
Source: https://huggingface.co/papers/2605.09530
Abstract
MemPrivacy enables privacy-preserving personalized memory in edge-cloud environments by using type-aware placeholders to protect sensitive data while maintaining semantic integrity for effective memory operations.
As LLM-powered agents are increasingly deployed inedge-cloud environments,personalized memoryhas become a key enabler of long-term adaptation and user-centric interaction. However, cloud-assisted memory management exposes sensitive user information, while existingprivacy protectionmethods 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 decouplingprivacy protectionfrom semantic destruction, MemPrivacy minimizes sensitive data exposure while retaining the information required for effectivememory formationand retrieval. We also construct MemPrivacy-Bench for systematic evaluation, a dataset covering 200 users and over 52k privacy instances, and introduce a four-levelprivacy taxonomyfor configurable protection policies. Experiments show that MemPrivacy achieves strong performance inprivacy information extraction, substantially surpassing strong general-purpose models such as GPT-5.2 and Gemini-3.1-Pro, while also reducinginference latency. Across multiple widely usedmemory systems, MemPrivacy limitsutility lossto within 1.6%, outperforming baseline masking strategies. Overall, MemPrivacy offers an effective balance betweenprivacy protectionandpersonalized memoryutility for edge-cloud agents, enabling secure, practical, and user-transparent deployment.
View arXiv pageView PDFGitHub31Add to collection
Get this paper in your agent:
hf papers read 2605\.09530
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper4
#### IAAR-Shanghai/MemPrivacy-1.7B-SFT Text Generation• 2B• Updatedabout 2 hours ago • 74 • 20
#### IAAR-Shanghai/MemPrivacy-1.7B-RL Text Generation• 2B• Updatedabout 2 hours ago • 52 • 10
#### IAAR-Shanghai/MemPrivacy-4B-SFT Text Generation• 196k• Updatedabout 2 hours ago • 51 • 9
#### IAAR-Shanghai/MemPrivacy-4B-RL Text Generation• 4B• Updatedabout 2 hours ago • 107 • 9
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.09530 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.09530 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Personalize-then-Store: Benchmarking and Learning Personalized Memory for Long-horizon Agents
This paper introduces PerMemBench, the first benchmark for evaluating personalized memory systems in LLM-based agents, and proposes a session-level storage gating framework that adapts memory policies to individual user contexts.
MemArena: An Ego-Centric Benchmark for On-Device Agentic Personal Memory Assistants at Scale
MemArena is a new ego-centric benchmark for evaluating on-device personal memory assistants, using a MASim agent simulator to generate multi-session conversational worlds and ground truth across recall, reasoning, and trustworthiness dimensions. Initial results show memory-backend choice often matters more than reader scale, and permission-aware access remains a universal challenge.
PLACEMEM: Toward a Compute-Aware Memory Plane for Lifelong Agents
PlaceMem proposes a compute-aware memory plane for lifelong agents, using versioned memory capsules that unify semantic content and reusable runtime state to enable correction-aware reuse and avoid redundant computation.
@omarsar0: // AutoMem // I quite like this idea of metamemory. (bookmark it) This new research from Stanford treats agent's memory…
This Stanford research paper introduces AutoMem, a framework that treats agent memory management as a trainable skill. By optimizing memory structure and proficiency separately, AutoMem improves base agent performance 2x-4x on long-horizon tasks, enabling a 32B open-weight model to compete with frontier systems like Claude Opus 4.5 and Gemini 3.1 Pro Thinking.
Memora: A Harmonic Memory Representation Balancing Abstraction and Specificity
Memora is a scalable memory system for AI agents that decouples storage from retrieval, achieving state-of-the-art performance on long-horizon tasks while using up to 98% fewer tokens. The research is published at ICML 2026.