LATIDIA · Ciberseguridad
TRIAGE-CPG: Un estudio intermodelos del triaje de «malware» estático presupuestado con grandes modelos lingüísticos
arXiv:2611.21331v1 Tipo de anuncio: nuevo Resumen: Al clasificar binarios desconocidos, los analistas de seguridad deben decidir qué muestras considerar benignas, marcar como maliciosas o derivar para un análisis más profundo. Grandes modelos de lenguaje (L
WhatsApp ↗Telegram ↗
La noticia
arXiv:2611.21331v1 Announce Type: new Abstract: When triaging unknown binaries, security analysts must decide which samples to clear as benign, flag as malicious, or escalate for deeper analysis. Large Language Models (LLMs) could support these decisions by interpreting decompiled code, but a binary can contain hundreds or thousands of functions, making exhaustive LLM analysis impractical. We study how current LLMs behave when static triage is formulated as budgeted evidence acquisition: which functions they inspect and how they trade off benign clearance, malware detection, and escalation. To enable a controlled comparison, we implement TRIAGE-CPG, a fixed pipeline over a code property graph (CPG) of a decompiled binary. An LLM-free heuristic selects seed