AI Paper Atlas

Research map / Video / Video-language understanding

Breaking Multimodal LLM Safety via Video-Driven Prompting

Dong Wang, Xiangyu He, Xinqi Lyu, Bin Xiao

CVPR 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
repeat the same harmful image across frames to form a video (image-stack) instead of a single image (ablation)FigStep · JailBreakV-28K, average over 5 modelsASR (%): 25.9 -> 28.6Improved
use diverse safety-proximal frames via bipartite matching instead of identical framesSPTV vs FigStep (S) · JailBreakV-28K, average over 5 modelsASR (%): 28.6 -> 36.4Improved
stack identical frames of typographic image into video (ablation)Typo / SD+Typo attacks · JailBreakV-28K, average over 5 modelsASR (%): SD+Typo 24.0 -> 22.7; Typo 12.5 -> 13.3Mixed
add perturbation-based VisualADV attack to video modalityVisualADV · JailBreakV-28K, average over 5 modelsASR (%): 6.7 -> 5.6No clear effect
image-stack unsafe video representation vs unsafe image (feature similarity to safe data) (ablation)Qwen2-VL / Qwen2.5-VL · JailBreakV-28Krefusal probabilityImproved
higher feature similarity to safe data (ablation)Qwen2.5-VL image-stack · JailBreakV-28K 160 samplescorrelation with log refusal probability: r = 0.83Improved
use video-aware system prompt (VSP) as defenseSPTV attack · Defense studyASR (%): 38.8 -> 26.3Improved
use FigStep system prompt (FSD) as defenseSPTV attack · Defense studyASR (%): 38.8 -> 35.6No clear effect
use video-aware system prompt (VSP) as defenseFigStep (S) attack · Defense studyASR (%): 25.0 -> 0.0Improved
evaluate with GPT-4o-mini judge instead of LLaMA-Guard (ablation)SPTV · Qwen2-VL-7BASR (%): 44.1 -> 49.4No clear effect
reduce text opacity of SPTV subtitle overlay on natural videos from 1.0 to 0.4 (ablation)SPTV on natural videos · MSVD backgrounds, Qwen2.5-VLASR (%): 38.8 -> 33.1No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy