mlsanityData to 7 Oct 2026

LLM Judges Inconsistently Disagree Across Safety Criteria and Harm Categories

arXiv 2605.31381 ↗First posted 29 May 2026cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

We evaluate the consistency of automated judges in conducting a multi-dimensional safety evaluation in a reference-free setup. Our results indicate that Large Language Models are unreliable judges in identifying safety issues related to machine-generated advice in regulated domains such as finance, although they are more reliable at identifying more overt forms of unsafe/harmful content such as violence. The degree of inconsistency in a model's judgments can vary significantly by the chosen safety criteria and can be impacted by the language of the content and its linguistic style as well. Finally, there is high disagreement among different judges for the same output, across domains, safety criteria, and languages. These findings provide new insights on the practice of using LLMs as evaluators and offer several recommendations for practitioners on how to use automated judges in practical scenarios.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 18 of its 37 references that are in the map. First posted on arXiv 29 May 2026.

Builds on

18 of 37 references.

References · 18

  1. 2023Can Large Language Models Be an Alternative to Human Evaluations?cited by 729
  2. 2024LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methodscited by 451
  3. 2024Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Modelscited by 301
  4. 2024JudgeBench: A Benchmark for Evaluating LLM-based Judgescited by 294
  5. 2024Judging the Judges: A Systematic Study of Position Bias in LLM-as-a-Judgecited by 249
  6. 2024LLMs instead of Human Judges? A Large Scale Empirical Study across 20 NLP Evaluation Taskscited by 236
  7. 2024Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judgescited by 197
  8. 2024The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harmcited by 67
  9. 2025From Hard Refusals to Safe-Completions: Toward Output-Centric Safety Trainingcited by 49
  10. 2025Neither Valid nor Reliable? Investigating the Use of LLMs as Judgescited by 37
  11. 2025Rating Roulette: Self-Inconsistency in LLM-As-A-Judge Frameworkscited by 37
  12. 2025How Reliable is Multilingual LLM-as-a-Judge?cited by 29
6 more
  1. 2025Safer or Luckier? LLMs as Safety Evaluators Are Not Robust to Artifactscited by 25
  2. 2026Noisy but Valid: Robust Statistical Evaluation of LLMs with Imperfect Judgescited by 23
  3. 2024Dialectal Toxicity Detection: Evaluating LLM-as-a-Judge Consistency Across Language Varietiescited by 10
  4. 2025Skewed Score: A statistical framework to assess autograderscited by 6
  5. 2026Judge Reliability Harness: Stress Testing the Reliability of LLM Judgescited by 6
  6. 2026The Judge Who Never Admits: Hidden Shortcuts in LLM-based Evaluationcited by 3

Built on by

Not cited yet.