SafeVid: Toward Safety Aligned Video Large Multimodal Models

As Video Large Multimodal Models (VLMMs) rapidly advance, their inherent complexity introduces significant safety challenges, particularly the issue of mismatched generalization where static safety alignments fail to transfer to dynamic video contexts. We introduce SafeVid, a framework designed to instill video-specific safety principles in VLMMs. SafeVid uniquely transfers robust textual safety alignment capabilities to the video domain by employing detailed textual video descriptions as an interpretive bridge, facilitating LLM-based rule-driven safety reasoning. This is achieved through a closed-loop system comprising: 1) generation of SafeVid-350K, a novel 350, 000-pair video-specific safety preference dataset; 2) targeted alignment of VLMMs using Direct Preference Optimization (DPO); and 3) comprehensive evaluation via our new SafeVidBench benchmark. Alignment with SafeVid-350K significantly enhances VLMM safety, with models like LLaVA-NeXT-Video demonstrating substantial improvements (e. g. , up to 42. 39%) on SafeVidBench. SafeVid provides critical resources and a structured approach, demonstrating that leveraging textual descriptions as a conduit for safety reasoning markedly improves the safety alignment of VLMMs in complex multimodal scenarios.

Authors

• Yixu Wang Fudan University Shanghai Artificial Intelligence Laboratory
• Jiaxin Song University of Illinois at Urbana-Champaign
• Yifeng Gao
• Xin Wang
• Yang Yao Tsinghua University
• Yan Teng Shanghai Artificial Intelligence Laboratory
• Xingjun Ma Shanghai Key Lab of Intell. Info. Processing, School of CS, Fudan University
• Yingchun Wang Shanghai Artificial Intelligence Laboratory
• Yu-Gang Jiang Shanghai Key Lab of Intell. Info. Processing, School of CS, Fudan University Shanghai Collaborative Innovation Center of Intelligent Visual Computing

Keywords

No keywords are indexed for this paper.