作者位次 #4/12ACL 2026Findings· Findings of ACL
Do VLM s Have a Moral Backbone? A Study on the Fragile Morality of Vision-Language Models
Zhining Liu · Tianyi Wang · Xiao Lin · Penghao Ouyang · Gaotang Li · Ze Yang · Hui Liu · Sumit Keswani · Vishwa Pardeshi · Huijun Zhao · Wei Fan · Hanghang Tong
摘要
Despite substantial efforts toward improving the moral alignment of Vision-Language Models (VLMs), it remains unclear whether their ethical judgments are stable in realistic settings. This work studies moral robustness in VLMs, defined as the ability to preserve moral judgments under textual and...