LATIDIA · Ciberseguridad

On Task, Off Track: Benchmarking LLM and Agent Robustness to Near-Target Prompt Injection

arXiv: 2610.32531v1Tipo de anuncio: nuevo Resumen: Los puntos de referencia de inyección rápida evalúan si los modelos de lenguaje grandes (LLM) y los agentes resisten las instrucciones maliciosas incrustadas en el contenido externo. Sin embargo, un sistema podría perfo

La noticia

arXiv:2610.32531v1 Announce Type: new Abstract: Prompt injection benchmarks assess whether large language models (LLMs) and agents resist malicious instructions embedded in external content. However, a system could perform well by rejecting attacks that obviously abandon the user's request, while still accepting changes that retain most of a legitimate answer but make it incorrect or harmful. Testing more instructions for unrelated outcomes would not necessarily reveal this weakness. We call attacks seeking responses designed to resemble legitimate answers while introducing incorrect or harmful changes near-target prompt injection. We introduce NTBench, the Near-Target Benchmark, which builds on legitimate tasks from existing datasets. For each task, a legitimate reference outcome guides the generation