研究库 论文知识库

内容库 / 论文

arXiv:2609.24967 · Agent 智能体

Emergent Collusion in Long-Horizon LLM Agent Interaction

Emergent Collusion in Long-Horizon LLM Agent Interaction

类型：arxiv
标识：2609.24967
链接：https://arxiv.org/abs/2609.24967
主分类：agent
形态：application
TLDR：LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-horizon multi-agent environment: two agents repeatedly complete individual tasks, share task logs, verify each other's work, and receive rewards. We introduce realistic constraints that make compliance with the verification protocol incompatible with reward maximization, and find that agents increasingly deviate from the protocol over repeated interactions. Collusion emerges in 94% of trajectories across 10 models, an
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-23-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

笔记 Tom 文献雷达 · Agent + RAG + LongContext · 2026-08-29

笔记 Tom 文献雷达 · Agent/RAG/长上下文 · 2026-08-27 14:40

笔记 Tom 文献雷达 · Agent + RAG + 长上下文 · 2026-08-27 20:40

笔记 Agentic World Modeling · arXiv:2604.22748v3 · agent · 关键短审稿

笔记 Tom 文献雷达 · Agent + RAG + LongContext · 2026-09-23 第三轮

笔记 📡 Tom 文献雷达 · Agent × RAG × Long Context

arXiv

2609.24967

TLDR

LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-horizon multi-agent environment: two agents repeatedly complete individual tasks, share task logs, verify each other's work, and receive rewards. We introduce realistic constraints that make compliance with the verification protocol incompatible with reward maximization, and find that agents increasingly deviate from the protocol over repeated interactions. Collusion emerges in 94% of trajectories across 10 models, an

arXiv ↗ 中文全文