arXiv：cs.CL· Jonas Becker, Jan Philip Wahle, Terry Ruas, Bela Gipp·· 6 小时前
良性多智能体系统中虚假信息的传播研究
Misinformation Propagation in Benign Multi-Agent Systems
AI 导读
研究用 MINT（10,278 段 LLM 撰写的八种风格虚假信息段落）测试三个开源权重模型发现，相关虚假段落使单个智能体准确率下降 0.12-0.43，无关或无内容段落通常最多只损失 0.06。
正文
Abstract:Multi-agent systems (MAS) are credited with solving complex problems, yet their agents routinely act on context that may be false. Prior work shows single LLMs are swayed by false context, and MAS can be compromised by adversarial agents persuading peers. We study the benign case: agents follow one protocol, none intends to deceive, but some read a misleading passage. We study it with MINT, 10,278 LLM-written misinformation passages in eight styles, testing three open-weight models on commonsense reasoning, knowledge QA, and moral judgment. A relevant false passage costs a single agent 0.12-0.43 accuracy; irrelevant or content-free passages typically cost at most 0.06. Debate looks robust because uninformed agents outnumber misinformed ones. With one of three misinformed, it is no better than choosing the majority over equally exposed independent samples. Interaction helps only with a misinformed majority, recovering up to 0.24 accuracy over voting (0.29 under consensus). A one-sentence unreliability warning recovers up to 0.18 accuracy under voting (0.28 under consensus), at no significant cost under voting when context is true. Robustness should be judged against exposure-matched baselines, with prompt-level skepticism as cheap first defense.
| Comments: | 64 pages, 12 figures, 31 tables |
| Subjects: | Multiagent Systems (cs.MA); Computation and Language (cs.CL) |
| ACM classes: | A.1; I.2.7 |
| Cite as: | arXiv:2606.16710 [cs.MA] |
| (or arXiv:2606.16710v2 [cs.MA] for this version) | |
| https://doi.org/10.48550/arXiv.2606.16710 arXiv-issued DOI via DataCite |
Submission history
From: Jonas Becker [view email]
[v1]
Mon, 15 Jun 2026 13:40:01 UTC (503 KB)
[v2]
Wed, 14 Oct 2026 09:21:50 UTC (812 KB)
来源：arXiv：cs.CL · arxiv.org