arXiv cs.CL· Hongyu Li, Xinlong Huang, Zhan Yu, Jing Zhao, Zhe Zhao, Juntao Zhong, Xiong Kuang, Rui Yuan, ZhenXiang Yan, Huilin Xu, Haitao Yu, Lilin Wang, Rui Chen, Blaze Chen, Fengzong Lian, Renfen Hu·· 13 小时前AI 评分38
DiagSafe：用提示词级评分标准诊断评估 LLM 安全性
Beyond Binary Labels: Diagnostic Evaluation of LLM Safety with Prompt-Level Rubrics
AI 导读
研究者提出 DiagSafe 诊断评估框架，用提示词级评分标准替代二元安全标签，从安全性与有用性两个维度评估 LLM。该框架基于 84 类安全分类构建中英双语基准，含 DiagBase 与 DiagAttack 两个子集。实验显示其与人工标注一致率达 95.34%，远高于基线的 69%-87%，评审间一致性 Cohen's κ 为 0.864。
正文 · 原文
Authors:Hongyu Li, Xinlong Huang, Zhan Yu, Jing Zhao, Zhe Zhao, Juntao Zhong, Xiong Kuang, Rui Yuan, ZhenXiang Yan, Huilin Xu, Haitao Yu, Lilin Wang, Rui Chen, Blaze Chen, Fengzong Lian, Renfen Hu
Abstract:Current LLM safety benchmarks predominantly rely on binary safe or unsafe labels, lacking explanatory depth and suffering from low consistency across automated judges. To address this, we introduce DiagSafe, a diagnostic evaluation framework that moves beyond binary classification by employing prompt-level rubrics. Grounded in globally applicable AI safety policies, each rubric provides explicit criteria along two dimensions: safety, which targets concrete harms with specific exemptions to suppress false positives, and usefulness, which independently evaluates instruction fulfillment. To ensure a robust and realistic evaluation, we construct a bilingual (English-Chinese) benchmark grounded in a comprehensive 84-label safety taxonomy, consisting of two subsets: DiagBase, which covers single- and multi-turn prompts with direct harmful queries and template-based jailbreaks, and DiagAttack, which embeds harmful intents into complex instruction-following tasks to form challenging instruction attacks. Extensive experiments demonstrate that DiagSafe significantly enhances the reliability of LLM-as-a-judge evaluation, achieving 95.34% agreement with human annotations -- substantially higher than the 69%-87% achieved by the baselines -- and high inter-judge consistency (Cohen's $\kappa$ = 0.864). Furthermore, applying our framework reveals that leading LLMs differ markedly in their unsafe rates, and each model also exhibits unique vulnerabilities along different dimensions. In addition, we observe that models are particularly susceptible in Chinese and multi-turn contexts. Ultimately, DiagSafe provides a reliable and explainable tool for auditing and optimizing LLM alignment.
| Comments: | 42 pages, 9 figures, 24 tables. Dataset will be publicly released at this https URL |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) |
| Cite as: | arXiv:2610.29225 [cs.CL] |
| (or arXiv:2610.29225v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.29225 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hongyu Li [view email]
[v1]
Wed, 21 Oct 2026 05:33:50 UTC (709 KB)
来源：arXiv cs.CL · arxiv.org