MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models

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 atthis https URL.

@article{xu2025_2503.14827,
  title={ MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models },
  author={ Chejian Xu and Jiawei Zhang and Zhaorun Chen and Chulin Xie and Mintong Kang and Yujin Potter and Zhun Wang and Zhuowen Yuan and Alexander Xiong and Zidi Xiong and Chenhui Zhang and Lingzhi Yuan and Yi Zeng and Peiyang Xu and Chengquan Guo and Andy Zhou and Jeffrey Ziwei Tan and Xuandong Zhao and Francesco Pinto and Zhen Xiang and Yu Gai and Zinan Lin and Dan Hendrycks and Bo Li and Dawn Song },
  journal={arXiv preprint arXiv:2503.14827},
  year={ 2025 }
}