[2402.04249] HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. arXiv Operational Status

[2402.04249] HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal Skip to main content arXiv is now an independent nonprofit! Learn more × Search arXiv Press Enter to search · Advanced search --> Computer Science > Machine Learning arXiv:2402.04249 (cs) [Submitted on 6 Feb 2024 ( v1 ), last revised 27 Feb 2024 (this version, v2)] Title: HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal Authors: Mantas Mazeika , Long Phan , Xuwang Yin , Andy Zou , Zifan Wang , Norman Mu , Elham Sakhaee , Nathaniel Li ,

related reading

• [2401.05566] Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingarxiv.org
• [2406.11717] Refusal in Language Models Is Mediated by a Single Directionarxiv.org
• [2407.15549] Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMsarxiv.org
• [2602.12316] GT-HarmBench: Benchmarking AI Safety Risks Through the Lens of Game Theoryarxiv.org
• [2502.05209] Model Tampering Attacks Enable More Rigorous Evaluations of LLM Capabilitiesarxiv.org
• [2302.10329] Harms from Increasingly Agentic Algorithmic Systemsarxiv.org
• [2406.04313] Improving Alignment and Robustness with Circuit Breakersarxiv.org
• [2304.03279] Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmarkarxiv.org
• [2406.04219] Multi-Agent Imitation Learning: Value is Easy, Regret is Hardarxiv.org
• [2211.14946] Self-Destructing Models: Increasing the Costs of Harmful Dual Uses of Foundation Modelsarxiv.org
• [2402.12875] Chain of Thought Empowers Transformers to Solve Inherently Serial Problemsarxiv.org
• [2312.15159] Understanding the Potential of FPGA-Based Spatial Acceleration for Large Language Model Inferencearxiv.org