LATIDIA · Ciberseguridad

Auditoría eficiente del comportamiento del agente de IA adversario a partir de los rastros del agente

arXiv:2610.07256v1 Tipo de anuncio: nuevo Resumen: los agentes de IA impulsados por modelos de lenguaje grandes (LLM) pueden realizar tareas complejas pero pueden dañar los sistemas en los que operan, ya sea intencional o involuntariamente. Agen existente

La noticia

arXiv:2610.07256v1 Announce Type: new Abstract: AI agents powered by large language models (LLMs) can perform complex tasks but may harm the systems they operate in, either intentionally or unintentionally. Existing agent monitoring approaches rely on rule-based guardrails or LLM-based trace auditing. However, rule-based guardrails can be bypassed through obfuscation and may miss harmful actions beyond their predefined rules, whereas applying an LLM to audit every action is costly. We present a two-stage agent trace auditing framework. The first stage uses single-event and trace-sequence rules to select pending actions for inspection; the second uses an LLM audit agent to examine each selected action in the context of the agent's preceding trace