Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition

2024/06/12 by Edoardo Debenedetti, Javier Rando, Debenedetti, Edoardo +41 · 1 voice · 9 citations
Computer Science · Materials Science · #Artificial Intelligence (cs.AI) #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Fusion materials and technologies #cs.AI #cs.CR

paper · pdf · doi:10.48550/arxiv.2406.07954

openalex publication_date 2024/06/12 · arxiv published 2024/06/12 · arxiv updated 2024/06/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Large language model systems face important 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.

Cited by

• ContextLeak: Auditing Leakage in Private In-Context Learning Methods
• Prompt Governance? On Governing Technologies Governed by Natural Language
• SnapAudit: Active Auditing of Differentially Private In-Context Learning via Snapshot-Based Simulation
• SoK: Systematizing LLM Prompt Security: Taxonomies, Datasets, and Unified Evaluation of Attacks and Defenses
• Exploiting Web Search Tools of AI Agents for Data Exfiltration
• LLMail-Inject: A Dataset from a Realistic Adaptive Prompt Injection Challenge
• Simple Prompt Injection Attacks Can Leak Personal Data Observed by LLM Agents During Task Execution
• ProxyPrompt: Securing System Prompts against Prompt Extraction Attacks
• Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks

Discussions

• 1) Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition. 📅 Fri 13 Dec 4:30 p.m. PST — 7:30 p.m. PST 📍 Spotlight Poster #5203 (West Ballroom A-D) arxiv.org/abs/2406.0795 [bsky, 1 points, 1 comments]

Related

• Emerging Security Challenges of Large Language Models
• A New Era in LLM Security: Exploring Security Concerns in Real-World LLM-based Systems
• Breaking to Build: A Threat Model of Prompt-Based Attacks for Securing LLMs
• Security Concerns for Large Language Models: A Survey
• Coercing LLMs to do and reveal (almost) anything