LATIDIA · Ciberseguridad

StateFence: Protección del Estado de la Tarea contra la Contaminación de la Cadena de Evidencia en Agentes LLM del Equipo Azul

arXiv:2611.02605v1 Anuncio Tipo: nuevo Resumen: La inyección indirecta de mensajes crea un riesgo distintivo para los agentes del modelo de lenguaje grande (LLM) del equipo azul en los Centros de Operaciones de Seguridad (SOC). Se pueden em instrucciones maliciosas

La noticia

arXiv:2611.02605v1 Announce Type: new Abstract: Indirect prompt injection creates a distinctive risk for blue-team Large Language Model (LLM) agents in Security Operations Centers (SOCs). Malicious instructions can be embedded in telemetry, tickets, retrieval results, or memory and then propagate through alert triage, incident summarization, response recommendation, and Indicator of Compromise (IOC) extraction as locally plausible intermediate decisions. We characterize this failure mode as evidence-chain contamination of protected task state. The main challenge is task dependent: a suspicious string may be malicious control text for summarization but essential evidence for IOC extraction. To address this challenge, we propose StateFence, a task-aware evaluation and defense framework that treats severity labels, incident narratives,