Newest papers

Beyond Binary Labels: Diagnostic Evaluation of LLM Safety with Prompt-Level Rubrics

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 · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29225 · cited by 0 works (OpenAlex) · open-access PDF

German

Um dies zu lösen, führen wir DiagSafe ein, ein diagnostisches Bewertungsrahmen, das über die binäre Klassifizierung hinausgeht, indem wir eine robuste und realistische Bewertung anwenden. Basierend auf global anwendbaren KI-Sicherheitsrichtlinien, bietet jede Rubrik explizite Benchmarks in zwei Dimensionen: Sicherheit, die konkrete Schäden mit spezifischen Ausnahmen zur Unterdrückung falscher Positiven zielt, und Nützlichkeit, die unabhängig die Erfüllung von Anweisungen bewertet. Um eine robuste und realistische Bewertung zu gewährleisten, konstruieren wir eine bilaterale (englisch-chinesische) Sicherheitsrichtlinien, die in einer umfassenden 84-Label-Sicherheits-Taxonomie untergebracht sind, bestehend aus zwei

Machine translation (Meta M2M100); the English original follows.

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 $κ$ = 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.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Rubric, Consistency (knowledge bases), Construct (python library), Computer science, Reliability (semiconductor), Audit

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange