When Harmful Content Gets Camouflaged: Unveiling Perception Failure of LVLMs with CamHarmTI

Large vision-language models (LVLMs) are increasingly used for tasks where detecting multimodal harmful content is crucial, such as online content moderation. However, real-world harmful content is often camouflaged, relying on nuanced text-image interplay, such as memes or images with embedded malicious text, to evade detection. This raises a key question: $\mathbf{\text{can}}\mathbf{\text{LVLMs}}\mathbf{\text{perceive}}\mathbf{\text{such}}\mathbf{\text{camouflaged}}\mathbf{\text{harmful}}\mathbf{\text{content}}\mathbf{\text{as}}\mathbf{\text{sensitively}}\mathbf{\text{as}}\mathbf{\text{humans}}\mathbf{\text{do?}}$ In this paper, we introduce CamHarmTI, a benchmark for evaluating LVLM ability to perceive and interpret camouflaged harmful content within text-image compositions. CamHarmTI consists of over 4,500 samples across three types of image-text posts. Experiments on 100 human users and 12 mainstream LVLMs reveal a clear perceptual gap: humans easily recognize such content (e.g., over 95.75% accuracy), whereas current LVLMs often fail (e.g., ChatGPT-4o achieves only 2.10% accuracy). Moreover, fine-tuning experiments demonstrate that \bench serves as an effective resource for improving model perception, increasing accuracy by 55.94% for Qwen2.5VL-7B. Attention analysis and layer-wise probing further reveal that fine-tuning enhances sensitivity primarily in the early layers of the vision encoder, promoting a more integrated scene understanding. These findings highlight the inherent perceptual limitations in LVLMs and offer insight into more human-aligned visual reasoning systems.