Hidden Coalitions in Multi-Agent AI: A Spectral Diagnostic from Internal Representations

TL;DR

This paper presents a spectral method to detect coalition structures in multi-agent AI systems by analyzing internal neural representations, aiding in understanding emergent group behaviors.

Contribution

Introduces a practical spectral partitioning approach using mutual information graphs to identify coalition boundaries from internal agent states.

Findings

Successfully recovers programmed coalition structures in reinforcement learning environments.

Identifies coalition structures and dynamic reassignments in large language models.

Reveals hierarchical and conflicting interaction patterns not detectable by scalar measures.

Abstract

Collections of interacting AI agents can form coalitions, creating emergent group-level organization that is critical for AI safety and alignment. However, observing agent behavior alone is often insufficient to distinguish genuine informational coupling from spurious similarity, as consequential coalitions may form at the level of internal representations before any overt behavioral change is apparent. Here, we introduce a practical method for detecting coalition structure from the internal neural representations of multi-agent systems. The approach constructs a pairwise mutual-information graph from the hidden states of agents and applies spectral partitioning to identify the most salient coalition boundary. We validate this method in two domains. First, in multi-agent reinforcement learning environments, the method successfully recovers programmed hierarchical and dynamic coalition structures and correctly rejects false positives arising from behavioral coordination without informational coupling. Second, using a large language model, the method identifies coalition structures implied by descriptive prompts, tracks dynamic team reassignments, and reveals a representational hierarchy where explicit labels dominate over conflicting interaction patterns. Across both settings, the recovered partition reveals subgroup organization that a scalar cross-agent mutual-information measure cannot distinguish. The results demonstrate that analyzing hidden-state mutual information through spectral partitioning provides a scalable diagnostic for identifying representational coalitions, offering a valuable tool for monitoring emergent structure in distributed AI systems.