Author name: Zhangyun Tan, Zeliang Zhang, Susan Liang, Yolo Yunlong Tang, Lisha Chen, Chenliang Xu

arXiv:2604.03114v1 Announce Type: new
Abstract: VLMs trained on web-scale data retain sensitive and copyrighted visual concepts that deployment may require removing. Training-based unlearning methods share a structural flaw: fine-tuning on a narrow fo…