SearchAttack: Red-Teaming LLMs against Knowledge-to-Action Threats under Online Web Search
SearchAttack: 对知识到行动威胁下对LLM进行红队攻击
机构 * Institute of Computing Technology, CAS(计算技术研究所,中国科学院) ; University of Chinese Academy of Sciences(中国科学院大学) ; People's Public Security University of China(中国人民公安大学) ; University of Science and Technology of China(中国科学技术大学) ; Guangdong Laboratory of Artificial Intelligence and Digital Economy(广东人工智能与数字经济实验室) ; Tsinghua University(清华大学)
AI总结 SearchAttack通过重新表述有害语义和测试奖励追逐偏差,揭示LLM在搜索增强任务中的安全漏洞。
Comments Misusing LLM-driven search for harmful information-seeking poses serious risks. We characterize its usability and impact through a comprehensive red-teaming and evaluation