LATIDIA · Ciberseguridad

Under the Skin: Infección por puerta trasera en modelos multimodales de lenguaje grande

arXiv: 2610.24058v1Tipo de Anuncio: nuevo Resumen: Los Modelos Multimodales de Lenguaje Grande (MLLM) han logrado un rendimiento impresionante en la generación de texto condicionado a entradas de diferentes modalidades. En este trabajo, presentamos

La noticia

arXiv:2610.24058v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) have achieved impressive performance in generating text conditioned on inputs from different modalities. In this work, we introduce backdoor infection, a cross-modal attack framework targeting MLLMs. Starting from a model already injected with a backdoor in one modality, our attack constructs a corresponding trigger in another modality, called the infected modality. Exploiting the model's shared latent space, the trigger in the infected modality reliably activates the malicious behavior, without modifying or poisoning the infected modality during training. We further develop a lightweight, constraint-aware optimization method for efficient and practical trigger generation. Extensive experiments across several state-of-the-art MLLMs, including LLaVA, InternVL,