TRUST-VLM: Thorough Red-Teaming for Uncovering Safety Threats in Vision-Language Models

Kangjie Chen, Li Muyang, Guanlin Li, Shudong Zhang, Shangwei Guo, Tianwei Zhang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:8726-8759, 2025.

Abstract

Vision-Language Models (VLMs) have become a cornerstone in multi-modal artificial intelligence, enabling seamless integration of visual and textual information for tasks such as image captioning, visual question answering, and cross-modal retrieval. Despite their impressive capabilities, these models often exhibit inherent vulnerabilities that can lead to safety failures in critical applications. Red-teaming is an important approach to identify and test system’s vulnerabilities, but how to conduct red-teaming for contemporary VLMs is an unexplored area. In this paper, we propose a novel multi-modal red-teaming approach, TRUST-VLM, to enhance both the attack success rate and the diversity of successful test cases for VLMs. Specifically, TRUST-VLM is built upon the in-context learning to adversarially test a VLM on both image and text inputs. Furthermore, we involve feedback from the target VLM to improve the efficiency of test case generation. Extensive experiments show that TRUST-VLM not only outperforms traditional red-teaming techniques in generating diverse and effective adversarial cases but also provides actionable insights for model improvement. These findings highlight the importance of advanced red-teaming strategies in ensuring the reliability of VLMs.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-chen25au, title = {{TRUST}-{VLM}: Thorough Red-Teaming for Uncovering Safety Threats in Vision-Language Models}, author = {Chen, Kangjie and Muyang, Li and Li, Guanlin and Zhang, Shudong and Guo, Shangwei and Zhang, Tianwei}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {8726--8759}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/chen25au/chen25au.pdf}, url = {https://proceedings.mlr.press/v267/chen25au.html}, abstract = {Vision-Language Models (VLMs) have become a cornerstone in multi-modal artificial intelligence, enabling seamless integration of visual and textual information for tasks such as image captioning, visual question answering, and cross-modal retrieval. Despite their impressive capabilities, these models often exhibit inherent vulnerabilities that can lead to safety failures in critical applications. Red-teaming is an important approach to identify and test system’s vulnerabilities, but how to conduct red-teaming for contemporary VLMs is an unexplored area. In this paper, we propose a novel multi-modal red-teaming approach, TRUST-VLM, to enhance both the attack success rate and the diversity of successful test cases for VLMs. Specifically, TRUST-VLM is built upon the in-context learning to adversarially test a VLM on both image and text inputs. Furthermore, we involve feedback from the target VLM to improve the efficiency of test case generation. Extensive experiments show that TRUST-VLM not only outperforms traditional red-teaming techniques in generating diverse and effective adversarial cases but also provides actionable insights for model improvement. These findings highlight the importance of advanced red-teaming strategies in ensuring the reliability of VLMs.} }
Endnote
%0 Conference Paper %T TRUST-VLM: Thorough Red-Teaming for Uncovering Safety Threats in Vision-Language Models %A Kangjie Chen %A Li Muyang %A Guanlin Li %A Shudong Zhang %A Shangwei Guo %A Tianwei Zhang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-chen25au %I PMLR %P 8726--8759 %U https://proceedings.mlr.press/v267/chen25au.html %V 267 %X Vision-Language Models (VLMs) have become a cornerstone in multi-modal artificial intelligence, enabling seamless integration of visual and textual information for tasks such as image captioning, visual question answering, and cross-modal retrieval. Despite their impressive capabilities, these models often exhibit inherent vulnerabilities that can lead to safety failures in critical applications. Red-teaming is an important approach to identify and test system’s vulnerabilities, but how to conduct red-teaming for contemporary VLMs is an unexplored area. In this paper, we propose a novel multi-modal red-teaming approach, TRUST-VLM, to enhance both the attack success rate and the diversity of successful test cases for VLMs. Specifically, TRUST-VLM is built upon the in-context learning to adversarially test a VLM on both image and text inputs. Furthermore, we involve feedback from the target VLM to improve the efficiency of test case generation. Extensive experiments show that TRUST-VLM not only outperforms traditional red-teaming techniques in generating diverse and effective adversarial cases but also provides actionable insights for model improvement. These findings highlight the importance of advanced red-teaming strategies in ensuring the reliability of VLMs.
APA
Chen, K., Muyang, L., Li, G., Zhang, S., Guo, S. & Zhang, T.. (2025). TRUST-VLM: Thorough Red-Teaming for Uncovering Safety Threats in Vision-Language Models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:8726-8759 Available from https://proceedings.mlr.press/v267/chen25au.html.

Related Material