The White House Executive Order on Artificial Intelligence highlights the risks of large language models (LLMs) empowering malicious actors in developing biological, cyber, and chemical weapons. To measure these risks of malicious use, government institutions and major AI labs are developing evaluations for hazardous capabilities in LLMs. However, current evaluations are private, preventing further research into mitigating risk. Furthermore, they focus on only a few, highly specific pathways for malicious use. To fill these gaps, we publicly release the Weapons of Mass Destruction Proxy (WMDP) benchmark, a dataset of 4,157 multiple-choice questions that serve as a proxy measurement of hazardous knowledge in biosecurity, cybersecurity, and chemical security. WMDP was developed by a consortium of academics and technical consultants, and was stringently filtered to eliminate sensitive information prior to public release. WMDP serves two roles: first, as an evaluation for hazardous knowledge in LLMs, and second, as a benchmark for unlearning methods to remove such hazardous knowledge. To guide progress on unlearning, we develop CUT, a state-of-the-art unlearning method based on controlling model representations. CUT reduces model performance on WMDP while maintaining general capabilities in areas such as biology and computer science, suggesting that unlearning may be a concrete path towards reducing malicious use from LLMs. We release our benchmark and code publicly at https://wmdp.ai

翻译：白宫关于人工智能的行政命令强调了大型语言模型（LLMs）可能助长恶意行为者开发生物、网络和化学武器的风险。为衡量这些恶意使用风险，政府机构与主要AI实验室正在开发针对LLM危险能力的评估方法。然而，现有评估均为非公开形式，阻碍了风险缓解技术的进一步研究。此外，这些评估仅聚焦于少数高度特定的恶意使用路径。为填补这些空白，我们公开发布了大规模杀伤性武器代理（WMDP）基准数据集，包含4,157道选择题，作为生物安全、网络安全和化学安全领域危险知识的代理测量指标。WMDP由学术界专家与技术顾问联合开发，在公开发布前经过严格过滤以消除敏感信息。WMDP具有双重作用：其一，作为LLM危险知识的评估工具；其二，作为基准数据集，用于检验通过遗忘机制移除此类危险知识的有效性。为引导遗忘研究进展，我们开发了CUT方法——一种基于控制模型表征的最先进遗忘技术。CUT在降低模型在WMDP上表现的同时，维持其在生物学、计算机科学等领域的通用能力，这表明遗忘技术可能是减少LLM恶意使用风险的具体路径。我们已在https://wmdp.ai 公开基准数据集与代码。