LATIDIA · Ciberseguridad

AgentCIScope: Evaluación de integridad contextual fundamentada para agentes de LLM

arXiv: 2610.19158v1Announce Type: cross Resumen: Los agentes de LLM son cada vez más capaces de manejar las tareas diarias del espacio de trabajo interactuando con entornos de aplicaciones a través de llamadas a herramientas. Sin embargo, a medida que los agentes leen un

La noticia

arXiv:2610.19158v1 Announce Type: cross Abstract: LLM agents are becoming increasingly capable of handling daily workspace tasks by interacting with application environments through tool calls. However, as agents read and write sensitive user data across email, documents, and other applications, a central privacy question is whether they preserve contextual integrity (CI), that is, whether the information flows they produce comply with context-specific disclosure constraints. Existing work either focuses on model-level privacy leakage or evaluates agents in limited application environments with manually constructed disclosure constraints. In this work, we introduce AgentCIScope, a unified platform and benchmark for evaluating agent-level contextual integrity through executable, multi-step tasks across diverse, realistic application environments. AgentCIScope transforms