arXiv 2609.123202026-09-14cs.AIcs.LG

AIM：面向多智能体多用户LLM系统的隐私感知可互操作记忆框架

AIM: A Privacy-Aware Interoperable Memory Framework for Multi-Agent Multi-User LLM Systems

Microsoft（微软）

机构由 AI 辅助整理，请以论文原文为准。

Zachary Johnson, Nigel Boachie Kumankumah, Somya Chatterjee, Tejas Sathyamurthi, Min Chen, Xinyi Alice Li, Xiao Wang, Emily Morgan Gelchie, Jessica Lin, Sadid A… 展开作者

Zachary Johnson, Nigel Boachie Kumankumah, Somya Chatterjee, Tejas Sathyamurthi, Min Chen, Xinyi Alice Li, Xiao Wang, Emily Morgan Gelchie, Jessica Lin, Sadid A. Hasan, Sulaiman Vesal

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对现有LLM记忆系统局限于单用户的问题，提出隐私感知的AIM框架，动态分类私有与公共记忆并实施索引级访问控制，配合首个多用户记忆基准MUMBench，实现高准确率的多操作记忆管理。

AI中文摘要：

传统的大语言模型（LLM）局限于单个用户会话，其知识仅限于单次对话，无法学习用户随时间演变的偏好。现有的智能体记忆系统解决了这一局限，但通常仅在单个用户层面运行，限制了可在用户间共享以改善下游响应的公共知识。我们提出了AIM（Agentic Interoperable Memory，智能体可互操作记忆），一个统一的、隐私感知的记忆框架，使多智能体、多用户LLM系统能够持久地管理私有记忆和共享记忆。AIM动态地将信息分类为私有（限于单个用户，其他用户不可访问）或公共（所有用户可访问）。它实施索引级访问控制，使得私有记忆仅能由其所有者检索，从而保护敏感数据，同时允许有益的共享知识改善协调性和一致性。我们还引入了MUMBench（Multi-User Memory Benchmark，多用户记忆基准），一个包含四个领域中包含私有和可共享信息的多用户交互数据集。据我们所知，MUMBench是首个旨在评估多用户环境中多种记忆操作（包括检索、创建、更新和删除）的公开数据集。在MUMBench上的三次独立运行中，AIM实现了96.0%的可见性分类准确率、58.8%的严格操作准确率和70.5%的状态感知操作准确率。

英文摘要：

Traditional large language models (LLMs) are scoped to individual user sessions, limiting their knowledge to a single conversation and preventing them from learning user preferences that evolve over time. Existing agentic memory systems address this limitation but generally operate at the individual-user level, restricting the public knowledge that could be shared across users to improve downstream responses. We introduce AIM (Agentic Interoperable Memory), a unified, privacy-aware memory framework that enables multi-agent, multi-user LLM systems to persistently manage private and shared memory. AIM dynamically classifies information as private, scoped to one user and inaccessible to others, or public, accessible to all users. It enforces index-level access controls so that private memories are retrievable only by their owner, protecting sensitive data while allowing beneficial shared knowledge to improve coordination and consistency. We also introduce MUMBench (Multi-User Memory Benchmark), a dataset of multi-user interactions containing private and shareable information across four domains. To our knowledge, MUMBench is the first public dataset designed to evaluate multiple memory operations, including retrieval, creation, update, and deletion, in a multi-user environment. Across three independent runs on MUMBench, AIM achieves 96.0% visibility classification accuracy, 58.8% strict operation accuracy, and 70.5% state-aware operation accuracy.