Misinformation Propagation in Benign Multi-Agent Systems

2026-06-15

AI Summary

This study investigates error propagation and system robustness in multi-agent systems operating in high-stakes scenarios, where misinformation received by a single agent can compromise overall reliability. By intentionally injecting errors into both single-agent and multi-agent settings, the authors examine how such errors propagate across reasoning, knowledge, and alignment tasks under varying decision protocols. Leveraging a large language model–based multi-agent debate framework, the work demonstrates that while debate alone cannot fully eliminate erroneous influences, multi-agent systems consistently outperform their single-agent counterparts. Notably, a majority of correctly informed agents can guide misled peers toward accurate conclusions, and consensus mechanisms prove more stable than voting under peer pressure. System robustness is shown to critically depend on both individual model capabilities and the method of information aggregation.

Abstract

Multi-agent systems, in which multiple large language model agents solve problems through turn-based interaction, are increasingly deployed in high-stakes settings such as medical diagnosis, legal analysis, and forensic decision-making. Their reliability can be at risk when single agents reason from incorrect or misleading context, e.g., from tool calls, since errors may propagate through agent interactions. This work studies this risk by injecting intent-based misinformation into benign single-agent and multi-agent systems across reasoning, knowledge, and alignment tasks. We find that misinformation can degrade single-agent performance and persists across multi-agent debate, with agents often retaining answers introduced by misinformed peers. Nevertheless, multi-agent debate reduces the resulting performance degradation compared to single-agent prompting, especially when most agents are not exposed to misinformation. Robustness depends on group composition and decision protocol. Consensus can be more stable than voting under peer pressure, while majorities can often steer misinformed agents back toward correct answers. Our results show that misinformation robustness in multi-agent systems depends on the underlying model and also on how agents exchange information and aggregate decisions.

Problem

• misinformation propagation
• multi-agent systems
• large language models
• reliability
• agent interaction

Innovation

• misinformation propagation
• multi-agent systems
• large language models
• agent debate
• decision aggregation