LATIDIA · Ciberseguridad

LITMUS: Benchmarking Behavioral Jailbreaks of LLM Agents in Real OS Environments

arXiv: 2605.10779v2Tipo de anuncio: reemplazar Resumen: La rápida proliferación de agentes autónomos basados en LLM en entornos de sistemas operativos reales introduce una nueva categoría de riesgo de seguridad más allá de la seguridad del contenido: comportamiento ja

La noticia

arXiv:2605.10779v2 Announce Type: replace Abstract: The rapid proliferation of LLM-based autonomous agents in real operating system environments introduces a new category of safety risk beyond content safety: behavior jailbreak, where an adversary induces an agent to execute dangerous OS-level operations with irreversible consequences. Existing benchmarks either evaluate safety at the semantic layer alone, missing physical-layer harms, or fail to isolate test cases, letting earlier runs contaminate later ones. We present LITMUS (LLM-agents In-OS Testing for Measuring Unsafe Subversion), a benchmark addressing both gaps via a semantic-physical dual verification mechanism and OS-level state rollback. LITMUS comprises 819 high-risk test cases organized into one harmful seed subset and six attack-extended subsets covering