It is shown that the diversity of red-teaming instances generated by ARMs is significantly higher, revealing emerging vulnerabilities in VLMs, and construct ARMs-Bench, a large-scale multimodal safety dataset comprising over 30K red-teaming instances spanning 51 diverse risk categories, grounded in both real-world mult...
Zhao-Run Chen, Xun Liu, Min-Tong Kang et al.· International Conference on...· 5 citations
This work provides the first systematic characterization for the effectiveness of different injection placements, spanning both coarse-grained environmental surfaces and fine-grained positions, and develops learning-based attackers that generate injection content and select better injection placements.
Cheng-Quan Guo, Xiao-Dong Liu, Wei-Wei Yang et al.· 0 citations
This work introduces AgentCIScope, a unified platform and benchmark for evaluating agent-level contextual integrity through executable, multi-step tasks across diverse, realistic application environments, and serves as a unified testbed for developing privacy-preserving agents.
Cheng-Quan Guo, Zhao-Run Chen, Zi-Nan Lin et al.· 0 citations
This work proposes Twin Agent, a general privilege separation design pattern inspired by residual coding in the agent context that preserves high task utility while preventing prompt injection attacks, outperforming both undefended agents and privilege separation baselines.
Zhan-Hao Hu, Dennis Jacob, Xiao Huang et al.· arXiv.org· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.