MAS-Shield: A Defense Framework for Secure and Efficient LLM MAS

arXiv:2511.22924v3 Announce Type: replace-cross Abstract: Large Language Model (LLM)-based Multi-Agent Systems (MAS) are susceptible to linguistic attacks that can trigger cascading failures across the network. Existing defenses face a fundamental dilemma: lightweight single-auditor methods are prone to single points of failure, while robust committee-based approaches incur prohibitive computational costs in multi-turn interactions. To address this challenge, we propose extbf{MAS-Shield}, a secure and efficient defense framework designed with a coarse-to-fine filtering pipeline. Rather than applying uniform scrutiny, MAS-Shield dynamically allocates defense resources through a three-stage protocol: (1) extbf{Critical Agent Selection } strategically targets high-influence nodes to narrow the defense surface; (2) extbf{Light Auditing} employs lightweight sentry models to rapidly filter the majority of benign cases; and (3) extbf{Global Consensus Auditing} escalates only suspicious or ambiguous signals to a heavyweight committee for definitive arbitration. This hierarchical design effectively optimizes the security-efficiency trade-off. Experiments demonstrate that MAS-Shield achieves a 92.5% recovery rate against diverse adversarial scenarios and reduces defense latency by over 70% compared to existing methods.

Source: arXiv cs.AI | 2026-09-18