1 paper · 1 filter
Youngjin Na, Sangheon Jeong, Youngwan Lee +3
With the growing deployment of Vision-Language Models (VLMs) in real-world applications, previously overlooked safety risks are becoming increasingly evident. In particular, seemin…