VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908380533096448 |
|---|---|
| author | Geng, Jiahui Li, Qing Chen, Zongxiong Wang, Yuxia Zhu, Derui Xie, Zhuohan Lyu, Chenyang Chen, Xiuying Nakov, Preslav Karray, Fakhri |
| author_facet | Geng, Jiahui Li, Qing Chen, Zongxiong Wang, Yuxia Zhu, Derui Xie, Zhuohan Lyu, Chenyang Chen, Xiuying Nakov, Preslav Karray, Fakhri |
| contents | The rapid advancement of vision-language models (VLMs) has brought a lot of attention to their safety alignment. However, existing methods have primarily focused on model undersafety, where the model responds to hazardous queries, while neglecting oversafety, where the model refuses to answer safe queries. In this paper, we introduce the concept of $\textit{safety calibration}$, which systematically addresses both undersafety and oversafety. Specifically, we present $\textbf{VSCBench}$, a novel dataset of 3,600 image-text pairs that are visually or textually similar but differ in terms of safety, which is designed to evaluate safety calibration across image-centric and text-centric scenarios. Based on our benchmark, we evaluate safety calibration across eleven widely used VLMs. Our extensive experiments revealed major issues with both undersafety and oversafety. We further investigated four approaches to improve the model's safety calibration. We found that even though some methods effectively calibrated the models' safety problems, these methods also lead to the degradation of models' utility. This trade-off underscores the urgent need for advanced calibration methods, and our benchmark provides a valuable tool for evaluating future approaches. Our code and data are available at https://github.com/jiahuigeng/VSCBench.git. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_20362 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration Geng, Jiahui Li, Qing Chen, Zongxiong Wang, Yuxia Zhu, Derui Xie, Zhuohan Lyu, Chenyang Chen, Xiuying Nakov, Preslav Karray, Fakhri Information Retrieval Artificial Intelligence The rapid advancement of vision-language models (VLMs) has brought a lot of attention to their safety alignment. However, existing methods have primarily focused on model undersafety, where the model responds to hazardous queries, while neglecting oversafety, where the model refuses to answer safe queries. In this paper, we introduce the concept of $\textit{safety calibration}$, which systematically addresses both undersafety and oversafety. Specifically, we present $\textbf{VSCBench}$, a novel dataset of 3,600 image-text pairs that are visually or textually similar but differ in terms of safety, which is designed to evaluate safety calibration across image-centric and text-centric scenarios. Based on our benchmark, we evaluate safety calibration across eleven widely used VLMs. Our extensive experiments revealed major issues with both undersafety and oversafety. We further investigated four approaches to improve the model's safety calibration. We found that even though some methods effectively calibrated the models' safety problems, these methods also lead to the degradation of models' utility. This trade-off underscores the urgent need for advanced calibration methods, and our benchmark provides a valuable tool for evaluating future approaches. Our code and data are available at https://github.com/jiahuigeng/VSCBench.git. |
| title | VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration |
| topic | Information Retrieval Artificial Intelligence |
| url | https://arxiv.org/abs/2505.20362 |