MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915204385734656 |
|---|---|
| author | Xu, Chejian Zhang, Jiawei Chen, Zhaorun Xie, Chulin Kang, Mintong Potter, Yujin Wang, Zhun Yuan, Zhuowen Xiong, Alexander Xiong, Zidi Zhang, Chenhui Yuan, Lingzhi Zeng, Yi Xu, Peiyang Guo, Chengquan Zhou, Andy Tan, Jeffrey Ziwei Zhao, Xuandong Pinto, Francesco Xiang, Zhen Gai, Yu Lin, Zinan Hendrycks, Dan Li, Bo Song, Dawn |
| author_facet | Xu, Chejian Zhang, Jiawei Chen, Zhaorun Xie, Chulin Kang, Mintong Potter, Yujin Wang, Zhun Yuan, Zhuowen Xiong, Alexander Xiong, Zidi Zhang, Chenhui Yuan, Lingzhi Zeng, Yi Xu, Peiyang Guo, Chengquan Zhou, Andy Tan, Jeffrey Ziwei Zhao, Xuandong Pinto, Francesco Xiang, Zhen Gai, Yu Lin, Zinan Hendrycks, Dan Li, Bo Song, Dawn |
| contents | Multimodal foundation models (MMFMs) play a crucial role in various applications, including autonomous driving, healthcare, and virtual assistants. However, several studies have revealed vulnerabilities in these models, such as generating unsafe content by text-to-image models. Existing benchmarks on multimodal models either predominantly assess the helpfulness of these models, or only focus on limited perspectives such as fairness and privacy. In this paper, we present the first unified platform, MMDT (Multimodal DecodingTrust), designed to provide a comprehensive safety and trustworthiness evaluation for MMFMs. Our platform assesses models from multiple perspectives, including safety, hallucination, fairness/bias, privacy, adversarial robustness, and out-of-distribution (OOD) generalization. We have designed various evaluation scenarios and red teaming algorithms under different tasks for each perspective to generate challenging data, forming a high-quality benchmark. We evaluate a range of multimodal models using MMDT, and our findings reveal a series of vulnerabilities and areas for improvement across these perspectives. This work introduces the first comprehensive and unique safety and trustworthiness evaluation platform for MMFMs, paving the way for developing safer and more reliable MMFMs and systems. Our platform and benchmark are available at https://mmdecodingtrust.github.io/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_14827 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models Xu, Chejian Zhang, Jiawei Chen, Zhaorun Xie, Chulin Kang, Mintong Potter, Yujin Wang, Zhun Yuan, Zhuowen Xiong, Alexander Xiong, Zidi Zhang, Chenhui Yuan, Lingzhi Zeng, Yi Xu, Peiyang Guo, Chengquan Zhou, Andy Tan, Jeffrey Ziwei Zhao, Xuandong Pinto, Francesco Xiang, Zhen Gai, Yu Lin, Zinan Hendrycks, Dan Li, Bo Song, Dawn Computation and Language Artificial Intelligence Cryptography and Security Multimodal foundation models (MMFMs) play a crucial role in various applications, including autonomous driving, healthcare, and virtual assistants. However, several studies have revealed vulnerabilities in these models, such as generating unsafe content by text-to-image models. Existing benchmarks on multimodal models either predominantly assess the helpfulness of these models, or only focus on limited perspectives such as fairness and privacy. In this paper, we present the first unified platform, MMDT (Multimodal DecodingTrust), designed to provide a comprehensive safety and trustworthiness evaluation for MMFMs. Our platform assesses models from multiple perspectives, including safety, hallucination, fairness/bias, privacy, adversarial robustness, and out-of-distribution (OOD) generalization. We have designed various evaluation scenarios and red teaming algorithms under different tasks for each perspective to generate challenging data, forming a high-quality benchmark. We evaluate a range of multimodal models using MMDT, and our findings reveal a series of vulnerabilities and areas for improvement across these perspectives. This work introduces the first comprehensive and unique safety and trustworthiness evaluation platform for MMFMs, paving the way for developing safer and more reliable MMFMs and systems. Our platform and benchmark are available at https://mmdecodingtrust.github.io/. |
| title | MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models |
| topic | Computation and Language Artificial Intelligence Cryptography and Security |
| url | https://arxiv.org/abs/2503.14827 |