1
Deceive, Detect, and Disclose: Large Language Models Play Mini-Mafia
让LLM在迷你“狼人杀”中玩出欺骗、检测与披露，为多智能体社会智能提供理论新视角。
arXiv:2509.23023v3 Announce Type: replace Abstract: Large language models are increasingly deployed in multi-agent settings whose outcomes hinge on so…