LATIDIA · Ciberseguridad

Propagación de confianza y contención estructural en oleoductos LLM multiagente

arXiv: 2609.17648v1Tipo de anuncio: nuevo Resumen: Los sistemas LLM multiagente automatizan cada vez más las tareas que involucran a agentes con diferentes niveles de privilegios, creando un riesgo de seguridad en el que un agente de bajo privilegio comprometido

La noticia

arXiv:2609.17648v1 Announce Type: new Abstract: Multi-agent LLM systems increasingly automate tasks involving agents with different levels of privilege, creating a security risk in which a compromised low-privilege agent can influence a higher-privilege agent and trigger an unauthorized action. We study attack propagation in a four-agent LangGraph pipeline comprising a Supervisor, Researcher, Validator, and Executor. We evaluate shared-memory poisoning and indirect prompt injection through a forged approval embedded in a retrieved document. We compare the Validator's judgment with an independent authorization layer using task-bound signed tokens and a separately verified policy oracle. Our contribution is an empirical study of attack propagation, a component-level ablation of the authorization boundary, and the Judgment