CIA: Inferring the Communication Topology from LLM-based Multi-Agent Systems

TL;DR

This paper introduces CIA, a novel black-box attack method that infers communication topologies in LLM-based multi-agent systems, exposing significant privacy vulnerabilities.

Contribution

The paper presents CIA, a new attack technique that constructs adversarial queries to uncover communication structures in MAS, highlighting privacy risks.

Findings

CIA achieves an average AUC of 0.87 in topology inference.

Peak AUC of up to 0.99 demonstrates high attack effectiveness.

Reveals substantial privacy risks in LLM-based MAS.

Abstract

LLM-based Multi-Agent Systems (MAS) have demonstrated remarkable capabilities in solving complex tasks. Central to MAS is the communication topology which governs how agents exchange information internally. Consequently, the security of communication topologies has attracted increasing attention. In this paper, we investigate a critical privacy risk: MAS communication topologies can be inferred under a restrictive black-box setting, exposing system vulnerabilities and posing significant intellectual property threats. To explore this risk, we propose Communication Inference Attack (CIA), a novel attack that constructs new adversarial queries to induce intermediate agents' reasoning outputs and models their semantic correlations through the proposed global bias disentanglement and LLM-guided weak supervision. Extensive experiments on MAS with optimized communication topologies demonstrate the effectiveness of CIA, achieving an average AUC of 0.87 and a peak AUC of up to 0.99, thereby revealing the substantial privacy risk in MAS.