← 研踪 PaperTrailCVF · CVPR 2026:cvf:f4e30629dd27b96f415d2a0db9266f5a · 记录 #1
SafeLogo: Turning Your Logos into Jailbreak Shields via Micro-Regional Adversarial Training
来源日期（仅到年份） 2026
来源记录日期 2026
雷达观测 2026-11-05 03:27
摘要
Recent Vision-Language Models (VLMs) have become increasingly susceptible to jailbreak attacks, where adversarial prompts exploit subtle manipulation to circumvent safety alignment.The diversity and adaptability of such jailbreakers necessitate a defense mechanism with strong generalization capability.However, fine-tuning large-scale VLMs is computationally expensive, and introducing excessive visual or textual defense prompts is impractical for preserving image realism and model usability.We propose SafeLogo, which tunes a logo-sized visual prompt into a universal shield against diverse jailbreak attacks through micro-regional adversarial training.We are the first to integrate min-max adversarial optimization into visual defense prompt generation.Specifically, in the outer loop, SafeLogo injects compact, bounded perturbations into extremely small image regions (\leq 2% pixel coverage), effectively preserving both visual fidelity and semantic consistency.Meanwhile, overcoming the limitations of prior defenses constrained to a single attack direction or fixed benign supervision, the inner loop dynamically generates and selects the strongest one from a variety of jailbreakers.Extensive experiments on LLaVA-1.5-13B,MiniGPT-4, and Qwen3-VL show that SafeLogo markedly lowers jailbreak success rates on MM-SafetyBench, VLGuard, and FigStep, while preserving benign performance on MM-Vet and MME.
论文发现出处
这些页面提供论文线索，不代表同行评审或质量认证。
观测到的版本
平台记录号代表本站观测到的元数据变化，不是作者修订号。来源日期仅到月份或年份时，按该月或该年首日存档。
记录 #1 · 2026 · 最新观测版本
SafeLogo: Turning Your Logos into Jailbreak Shields via Micro-Regional Adversarial Training
Recent Vision-Language Models (VLMs) have become increasingly susceptible to jailbreak attacks, where adversarial prompts exploit subtle manipulation to circumvent safety alignment.The diversity and adaptability of such jailbreakers necessitate a defense mechanism with strong generalization capability.However, fine-tuning large-scale VLMs is computationally expensive, and introducing excessive visual or textual defense prompts is impractical for preserving image realism and model usability.We propose SafeLogo, which tunes a logo-sized visual prompt into a universal shield against diverse jailbreak attacks through micro-regional adversarial training.We are the first to integrate min-max adversarial optimization into visual defense prompt generation.Specifically, in the outer loop, SafeLogo injects compact, bounded perturbations into extremely small image regions (\leq 2% pixel coverage), effectively preserving both visual fidelity and semantic consistency.Meanwhile, overcoming the limitations of prior defenses constrained to a single attack direction or fixed benign supervision, the inner loop dynamically generates and selects the strongest one from a variety of jailbreakers.Extensive experiments on LLaVA-1.5-13B,MiniGPT-4, and Qwen3-VL show that SafeLogo markedly lowers jailbreak success rates on MM-SafetyBench, VLGuard, and FigStep, while preserving benign performance on MM-Vet and MME.
查看此版本解读 →查看此版本原文 ↗