LATIDIA · Ciberseguridad

El mismo cero: por qué ASR idéntico puede implicar diferentes garantías en LLM-Agent Security

arXiv:2610.04504v1 Tipo de anuncio: nuevo Resumen: la seguridad del agente LLM ha producido un denso panorama de defensas: refuerzo rápido, filtros de contenido, puertas de permisos, sandboxes, pero ningún marco le dice a un implementador qué

La noticia

arXiv:2610.04504v1 Announce Type: new Abstract: LLM-agent security has produced a dense landscape of defenses - prompt hardening, content filters, permission gates, sandboxes - yet no framework tells a deployer what a defense actually guarantees, or where that guarantee comes from. We apply Verification Autonomy Levels (VAL) - L0: LLM self-declaration; L1: deterministic rules; L2: objective ground truth; L3/L4: decidable completeness; L5: impossible - to 22 agent-security defenses; the taxonomy is falsifiable (10/10 prediction hits on frozen cards, flagged). We run the first controlled deployment-value comparison: at equal budget, a VAL-guided stack (confirmation gate + schema sandbox) versus a mainstream intuition stack (prompt hardening + keyword filter), 50 scenarios, 12 attack