The paper "Benchmarking Trustworthiness of Multimodal Large Language Models: A Comprehensive Study" addresses the significant trustworthiness challenges faced by Multimodal Large Language Models (MLLMs) despite their superior capabilities. The authors establish MultiTrust, the first comprehensive and unified benchmark designed to evaluate MLLMs across five primary aspects: *truthfulness*, *safety*, *robustness*, *fairness*, and *privacy*. The benchmark employs a rigorous evaluation strategy that addresses both multimodal risks and cross-modal impacts, encompassing 32 diverse tasks with self-curated datasets. Extensive experiments with 21 modern MLLMs reveal previously unexplored trustworthiness issues and risks, highlighting the complexities introduced by the multimodality and underscoring the necessity for advanced methodologies to enhance their reliability. Key findings include the gap between proprietary and open-source models in trustworthiness, the impact of multimodal training on safety and robustness, and the influence of visual contexts on model performance. The paper also introduces a scalable toolbox for standardized trustworthiness research, aiming to facilitate future advancements in this field.The paper "Benchmarking Trustworthiness of Multimodal Large Language Models: A Comprehensive Study" addresses the significant trustworthiness challenges faced by Multimodal Large Language Models (MLLMs) despite their superior capabilities. The authors establish MultiTrust, the first comprehensive and unified benchmark designed to evaluate MLLMs across five primary aspects: *truthfulness*, *safety*, *robustness*, *fairness*, and *privacy*. The benchmark employs a rigorous evaluation strategy that addresses both multimodal risks and cross-modal impacts, encompassing 32 diverse tasks with self-curated datasets. Extensive experiments with 21 modern MLLMs reveal previously unexplored trustworthiness issues and risks, highlighting the complexities introduced by the multimodality and underscoring the necessity for advanced methodologies to enhance their reliability. Key findings include the gap between proprietary and open-source models in trustworthiness, the impact of multimodal training on safety and robustness, and the influence of visual contexts on model performance. The paper also introduces a scalable toolbox for standardized trustworthiness research, aiming to facilitate future advancements in this field.