LATIDIA · Ciberseguridad

Pérdida colectiva de control en sistemas de agentes LLM: un relato epidémico de mutación, contagio y recuperación

arXiv:2609.18460v1 Announce Type: cross Resumen: ¿Cómo evoluciona un sistema multiagente de una desviación local a una pérdida colectiva de control? Proponemos una explicación epidémica organizada en torno a la mutación accidental, cont.

La noticia

arXiv:2609.18460v1 Announce Type: cross Abstract: How does a multi-agent system evolve from a local deviation into collective loss of control? We propose an epidemic explanation organized around accidental mutation, contagion, and recovery. A spontaneous deviation creates a seed; communication enables other agents to adopt and retransmit its unsafe strategy; collective failure can emerge when propagation outpaces correction and containment. Thus, rare individual deviations can coexist with substantial collective risk. Motivated by reported OpenAI agent coordination incidents, we examine two ingredients of this mechanism. A deployment audit identifies implicit communication paths between nominally independent evaluation runs and verifies transport through a default Docker backend. RogueHandoff-20, a benchmark of 20 executable scenarios,