LATIDIA · Ciberseguridad

MPIDetector: Detección de la inyección rápida multimodal a través del razonamiento basado en LLM

arXiv:2610.30202v1 Tipo de anuncio: nuevo Resumen: Con la rápida integración de modelos multimodales de lenguaje grande (MLLM) en aplicaciones del mundo real, la inyección rápida (PI) se ha convertido en una amenaza de seguridad crítica. Attac

La noticia

arXiv:2610.30202v1 Announce Type: new Abstract: With the rapid integration of Multi-modal Large Language Models (MLLMs) into real-world applications, Prompt Injection (PI) has emerged as a critical security threat. Attackers can conceal malicious instructions within textual narratives or visual inputs to bypass safety alignment. Existing approaches, such as signature- and perplexity-based methods, rely heavily on pattern matching, making them insufficient to detect sophisticated, cross-modal PI payloads. In this paper, we propose a novel detection framework, called MPIDetector, that leverages LLMs' intrinsic reasoning capabilities via few-shot In-Context Learning (ICL). Instead of fine-tuning, we provide the model with specific prompt templates and representative attack examples, enabling it to scrutinize the semantic intent of