VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration
Published in ACL 2025 Findings, 2025
VSCBench provides a systematic framework for evaluating and improving safety calibration in vision-language models, addressing critical gaps in current multimodal AI safety assessment.
Recommended citation: J Geng, Q Li, Z Chen, Y Wang, D Zhu, Z Xie, C Lyu, X Chen, P Nakov, et al. (2025). "VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration." ACL 2025 Findings.
Download Paper