Typographic attacks pose a critical threat to vision-language models (VLMs) by injecting misleading text into images and causing models to rely on adversarial textual cues rather than visual evidence. Existing defenses often require model-specific modifications, additional training, or access to internal model componen...
It is revealed that contexts with stronger semantic-shift capabilities are more likely to guide models toward recovering harmful meanings and achieving successful jailbreaks, and a black-box context-aware semantic-shift jailbreak framework with Iterative Context Optimization is proposed.
Hujian Zhu, Yi-Hao Huang, Felix Juefei-Xu et al.· 0 citations
This paper proposes a two-stage black-box framework to uncover and mitigate Positional Blind Spots, and evaluates its framework on five state-of-the-art VLA policies across two benchmarks, and finds that PBS are pervasive and spatially concentrated in all of them.
Dongdong An, Peng-Jie Zhao, Yi-Hao Huang et al.· 1 citation
Large vision-language models (LVLMs) have achieved remarkable progress in video understanding and reasoning. Despite extensive studies on text- and image-based jailbreaks, video jailbreaks against LVLMs remain largely unexplored. Existing video jailbreak methods mainly manipulate textual content embedded in videos, whi...
Ling Zhou, Yi-Hao Huang, Jing-Lin Sun et al.· 0 citations
Lingfuzz is a linguistic-aware fuzzing framework for continuing red teaming LLMs that enables generating diverse malicious instructions due to the unlimited space of lexical and syntactic choices, while having a continued testing capability by iteratively mutating the mutants.
Shuai Yuan, Nian Luo, Jingling Sun et al.· Proceedings of the ACM on So...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
