Emergent Social Intelligence Risks in Generative Multi-Agent Systems
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914444571836416 |
|---|---|
| author | Huang, Yue Jiang, Yu Wang, Wenjie Zhuang, Haomin Luo, Xiaonan Ma, Yuchen Xu, Zhangchen Chen, Zichen Moniz, Nuno Lin, Zinan Chen, Pin-Yu Chawla, Nitesh V Dziri, Nouha Sun, Huan Zhang, Xiangliang |
| author_facet | Huang, Yue Jiang, Yu Wang, Wenjie Zhuang, Haomin Luo, Xiaonan Ma, Yuchen Xu, Zhangchen Chen, Zichen Moniz, Nuno Lin, Zinan Chen, Pin-Yu Chawla, Nitesh V Dziri, Nouha Sun, Huan Zhang, Xiangliang |
| contents | Multi-agent systems composed of large generative models are rapidly moving from laboratory prototypes to real-world deployments, where they jointly plan, negotiate, and allocate shared resources to solve complex tasks. While such systems promise unprecedented scalability and autonomy, their collective interaction also gives rise to failure modes that cannot be reduced to individual agents. Understanding these emergent risks is therefore critical. Here, we present a pioneer study of such emergent multi-agent risk in workflows that involve competition over shared resources (e.g., computing resources or market share), sequential handoff collaboration (where downstream agents see only predecessor outputs), collective decision aggregation, and others. Across these settings, we observe that such group behaviors arise frequently across repeated trials and a wide range of interaction conditions, rather than as rare or pathological cases. In particular, phenomena such as collusion-like coordination and conformity emerge with non-trivial frequency under realistic resource constraints, communication protocols, and role assignments, mirroring well-known pathologies in human societies despite no explicit instruction. Moreover, these risks cannot be prevented by existing agent-level safeguards alone. These findings expose the dark side of intelligent multi-agent systems: a social intelligence risk where agent collectives, despite no instruction to do so, spontaneously reproduce familiar failure patterns from human societies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_27771 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Emergent Social Intelligence Risks in Generative Multi-Agent Systems Huang, Yue Jiang, Yu Wang, Wenjie Zhuang, Haomin Luo, Xiaonan Ma, Yuchen Xu, Zhangchen Chen, Zichen Moniz, Nuno Lin, Zinan Chen, Pin-Yu Chawla, Nitesh V Dziri, Nouha Sun, Huan Zhang, Xiangliang Multiagent Systems Computation and Language Computers and Society Multi-agent systems composed of large generative models are rapidly moving from laboratory prototypes to real-world deployments, where they jointly plan, negotiate, and allocate shared resources to solve complex tasks. While such systems promise unprecedented scalability and autonomy, their collective interaction also gives rise to failure modes that cannot be reduced to individual agents. Understanding these emergent risks is therefore critical. Here, we present a pioneer study of such emergent multi-agent risk in workflows that involve competition over shared resources (e.g., computing resources or market share), sequential handoff collaboration (where downstream agents see only predecessor outputs), collective decision aggregation, and others. Across these settings, we observe that such group behaviors arise frequently across repeated trials and a wide range of interaction conditions, rather than as rare or pathological cases. In particular, phenomena such as collusion-like coordination and conformity emerge with non-trivial frequency under realistic resource constraints, communication protocols, and role assignments, mirroring well-known pathologies in human societies despite no explicit instruction. Moreover, these risks cannot be prevented by existing agent-level safeguards alone. These findings expose the dark side of intelligent multi-agent systems: a social intelligence risk where agent collectives, despite no instruction to do so, spontaneously reproduce familiar failure patterns from human societies. |
| title | Emergent Social Intelligence Risks in Generative Multi-Agent Systems |
| topic | Multiagent Systems Computation and Language Computers and Society |
| url | https://arxiv.org/abs/2603.27771 |