Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914606243381248 |
|---|---|
| author | Nakamura, Mason Kumar, Abhinav Das, Saswat Abdelnabi, Sahar Mahmud, Saaduddin Fioretto, Ferdinando Zilberstein, Shlomo Bagdasarian, Eugene |
| author_facet | Nakamura, Mason Kumar, Abhinav Das, Saswat Abdelnabi, Sahar Mahmud, Saaduddin Fioretto, Ferdinando Zilberstein, Shlomo Bagdasarian, Eugene |
| contents | Multi-agent systems, where LLM agents communicate through free-form language, enable sophisticated coordination for solving complex cooperative tasks. This surfaces a unique safety problem when a group of agents forms a coalition and colludes to pursue secondary goals and degrade the joint objective. In this paper, we present Colosseum, a framework for auditing LLM agents' collusive behavior in multi-agent settings. We ground how agents cooperate through a formal multi-agent decision-making framework and measure action-based collusive behavior in actions via regret relative to the cooperative optimum and compare it with communication-based collusive behavior. Colosseum enables audits of LLM agents for collusion under benign settings, different coalition objectives, persuasion tactics, and network topologies. We then introduce a new behavioral probe by creating secret communication channels between agents, showing that most out-of-the-box models exhibit a propensity to collude under this probe, which we term emergent collusion. Furthermore, we discover ``collusion on paper'' when agents plan to collude in text but often pick non-collusive actions. Colosseum provides a new way to audit collusion in cooperative multi-agent systems while presenting observations about how collusion emerges, what affects collusion efficacy, and which strategies may mitigate it. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_15198 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems Nakamura, Mason Kumar, Abhinav Das, Saswat Abdelnabi, Sahar Mahmud, Saaduddin Fioretto, Ferdinando Zilberstein, Shlomo Bagdasarian, Eugene Multiagent Systems Artificial Intelligence Computation and Language Multi-agent systems, where LLM agents communicate through free-form language, enable sophisticated coordination for solving complex cooperative tasks. This surfaces a unique safety problem when a group of agents forms a coalition and colludes to pursue secondary goals and degrade the joint objective. In this paper, we present Colosseum, a framework for auditing LLM agents' collusive behavior in multi-agent settings. We ground how agents cooperate through a formal multi-agent decision-making framework and measure action-based collusive behavior in actions via regret relative to the cooperative optimum and compare it with communication-based collusive behavior. Colosseum enables audits of LLM agents for collusion under benign settings, different coalition objectives, persuasion tactics, and network topologies. We then introduce a new behavioral probe by creating secret communication channels between agents, showing that most out-of-the-box models exhibit a propensity to collude under this probe, which we term emergent collusion. Furthermore, we discover ``collusion on paper'' when agents plan to collude in text but often pick non-collusive actions. Colosseum provides a new way to audit collusion in cooperative multi-agent systems while presenting observations about how collusion emerges, what affects collusion efficacy, and which strategies may mitigate it. |
| title | Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems |
| topic | Multiagent Systems Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2602.15198 |