LATIDIA · Ciberseguridad
Proteger el reino de los agentes basados en habilidades en la dinastía de la IA computacional: integridad del flujo de control para las habilidades de los agentes
arXiv:2611.24185v1 Tipo de anuncio: nuevo Resumen: La integridad del flujo de control (CFI) limita un programa a un grafo de flujo de control establecido de antemano. Las habilidades de los agentes procedimentales trasladan este flujo de control al texto de la indicación, y un gran modelo lingüístico (LLM)
WhatsApp ↗Telegram ↗
La noticia
arXiv:2611.24185v1 Announce Type: new Abstract: Control-flow integrity (CFI) confines a program to a control-flow graph fixed ahead of time. Procedural agent skills move this control flow into prompt text, and a large language model (LLM) agent decides at runtime whether each step runs in its reasoning, a tool call, or a script. The procedure's control skeleton survives, but its enforceability depends on where each step runs. Many violations, such as overwriting a file before validation, use only operations the skill requires, so permission allowlists admit them. A deviating or prompt-injected model can also evade a monitor by regime migration, for example by claiming that a check passed. We present SkillCFI,