Vision-Language Models

MSTS: A Multimodal Safety Test Suite for Vision-Language Models

MSTS evaluates safety risks that emerge only when vision-language models interpret images and text together.