Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition
ETH Zurich Google · ETH Zurich · West University in Timisoara · ETHZ - ETH Zurich · New York University · Technion - Israel Institute of Technology, Technion · Microsoft Corp · CISPA Helmholtz Center for Information Security · Microsoft · Research, Microsoft · Microsoft Research · Robotic Systems Lab, ETH Zurich · ZHdK - Zurich University of the Arts
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
Large language model systems face significant security risks from maliciously crafted messages that aim to overwrite the system's original instructions or leak private data. To study this problem, we organized a capture-the-flag competition at IEEE SaTML 2024, where the flag is a secret string in the LLM system prompt. The competition was organized in two phases. In the first phase, teams developed defenses to prevent the model from leaking the secret. During the second phase, teams were challenged to extract the secrets hidden for defenses proposed by the other teams. This report summarizes the main insights from the competition. Notably, we found that all defenses were bypassed at least once, highlighting the difficulty of designing a successful defense and the necessity for additional research to protect LLM systems. To foster future research in this direction, we compiled a dataset with over 137k multi-turn attack chats and open-sourced the platform.