Benchmarking the Trustworthiness in Multimodal LLMs for Video Understanding

TL;DR

This paper introduces Trust-videoLLMs, a comprehensive benchmark for evaluating the trustworthiness of multimodal large language models in video understanding, focusing on accuracy, safety, fairness, and privacy across diverse tasks.

Contribution

It presents the first extensive benchmark assessing 23 videoLLMs on multiple trustworthiness dimensions, highlighting current limitations and guiding future improvements.

Findings

Significant gaps in dynamic scene understanding and robustness.

Open-source models sometimes outperform proprietary ones.

Scaling models does not always enhance trustworthiness.

Abstract

Recent advancements in multimodal large language models for video understanding (videoLLMs) have enhanced their capacity to process complex spatiotemporal data. However, challenges such as factual inaccuracies, harmful content, biases, hallucinations, and privacy risks compromise their reliability. This study introduces Trust-videoLLMs, a first comprehensive benchmark evaluating 23 state-of-the-art videoLLMs (5 commercial, 18 open-source) across five critical dimensions: truthfulness, robustness, safety, fairness, and privacy. Comprising 30 tasks with adapted, synthetic, and annotated videos, the framework assesses spatiotemporal risks, temporal consistency and cross-modal impact. Results reveal significant limitations in dynamic scene comprehension, cross-modal perturbation resilience and real-world risk mitigation. While open-source models occasionally outperform, proprietary models generally exhibit superior credibility, though scaling does not consistently improve performance. These findings underscore the need for enhanced training datat diversity and robust multimodal alignment. Trust-videoLLMs provides a publicly available, extensible toolkit for standardized trustworthiness assessments, addressing the critical gap between accuracy-focused benchmarks and demands for robustness, safety, fairness, and privacy.