arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2605.22786cs.AIcs.ETcs.LGcs.MA
收藏

LCGuard: 多智能体系统中安全KV共享的潜在通信守护者

LCGuard: Latent Communication Guard for Safe KV Sharing in Multi-Agent Systems

Rensselaer Polytechnic Institute(伦斯勒理工学院);IBM Research(IBM研究院)

Sadia Asif, Mohammad Mohammadi Amiri, Momin Abbas, Prasanna Sattigeri, Karthikeyan Natesan Ramamurthy

AI总结:

本文提出LCGuard框架,通过在智能体间共享KV缓存前学习表示层面的转换,以防止敏感信息泄露,同时在多个模型家族和多智能体基准测试中验证了其在减少重建攻击成功率和保持任务性能方面的有效性。

摘要:

基于大型语言模型(LLM)的多智能体系统越来越多地依赖中间通信来协调复杂任务。尽管大多数现有系统通过自然语言进行通信,但最近的研究表明,通过transformer键值(KV)缓存进行的潜在通信可以提高效率并保留更丰富的任务相关信息。然而,KV缓存也编码了上下文输入、中间推理状态和智能体特定信息,从而创建了一个可能传播敏感内容的不透明通道,而无需显式文本披露。为此,我们引入了LCGuard(潜在通信守护者),一个用于多智能体LLM系统中安全KV基于潜在通信的框架。LCGuard将共享的KV缓存视为潜在的工作记忆,并在缓存艺术制品传输到智能体之前学习表示层面的转换。我们通过重建正式化表示层面的敏感信息泄露操作:如果一个对抗性解码器可以从共享缓存艺术制品中恢复出智能体特定的敏感输入,则该共享缓存艺术制品是不安全的。这导致了一种对抗性训练公式,其中对抗者学习重建敏感输入,而LCGuard学习转换以保留任务相关语义并减少可重建的信息。在多个模型家族和多智能体基准测试中的实证评估表明,LCGuard在减少基于重建的泄露和攻击成功率的同时,能够保持与标准KV共享基线相比具有竞争力的任务性能。

英文摘要:

Large language model (LLM)-based multi-agent systems increasingly rely on intermediate communication to coordinate complex tasks. While most existing systems communicate through natural language, recent work shows that latent communication, particularly through transformer key-value (KV) caches, can improve efficiency and preserve richer task-relevant information. However, KV caches also encode contextual inputs, intermediate reasoning states, and agent-specific information, creating an opaque channel through which sensitive content may propagate across agents without explicit textual disclosure. To address this, we introduce \textbf{LCGuard} (Latent Communication Guard), a framework for safe KV-based latent communication in multi-agent LLM systems. LCGuard treats shared KV caches as latent working memory and learns representation-level transformations before cache artifacts are transmitted across agents. We formalize representation-level sensitive information leakage operationally through reconstruction: a shared cache artifact is unsafe if an adversarial decoder can recover agent-specific sensitive inputs from it. This leads to an adversarial training formulation in which the adversary learns to reconstruct sensitive inputs, while LCGuard learns transformations that preserve task-relevant semantics and reduce reconstructable information. Empirical evaluations across multiple model families and multi-agent benchmarks show that LCGuard consistently reduces reconstruction-based leakage and attack success rates while maintaining competitive task performance compared to standard KV-sharing baselines.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑