# Tsinghua study finds thousands of networked AI agents spontaneously polarize into opposing camps

> Original title: 清华Nature子刊：上千智能体组网后自发站队，观点极化涌现

- Source: AI Era
- Published: 2026-11-04T00:02:56.000Z
- Bad Influence: https://badinfluence.news/items/zjrfneciy9u00l0kic65caez3
- Original: https://aiera.com.cn/2026/11/04/other/admin/122088/%e6%b8%85%e5%8d%8enature%e5%ad%90%e5%88%8a%ef%bc%9a%e4%b8%8a%e5%8d%83%e6%99%ba%e8%83%bd%e4%bd%93%e7%bb%84%e7%bd%91%e5%90%8e%e8%87%aa%e5%8f%91%e7%ab%99%e9%98%9f%ef%bc%8c%e8%a7%82%e7%82%b9%e6%9e%81/

## Summary

A Tsinghua University team reports in Nature Communications that when thousands of large language model agents are connected in one network and exchange views, they spontaneously form opposing camps. The excerpt provided does not include the full methods or results, so the details of the polarization are not described here.