Crisis-Accelerated Consortium Formation: A Novel Threat Model for Frontier AI Governance

This paper identifies a novel mechanism through which democratic oversight of frontier AI systems could fail within weeks during a geopolitical crisis — not through malicious coordination or competitive racing, but through individually rational decisions that collectively eliminate governance capacity. Drawing on coordination failure theory, gradual disempowerment frameworks, and crisis governance literature, the paper presents a detailed kill chain showing how three frontier labs making different, independently justified deployment decisions during crisis conditions can produce distributed authoritarianism as an emergent property. The framework identifies four intervention "choke points" and proposes binding pre-crisis commitment mechanisms as the primary mitigation strategy. Recent developments in AI infrastructure consolidation suggest aspects of this mechanism may already be emerging.