RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models

• URL: http://arxiv.org/abs/2510.10390v1
• Date: Sun, 12 Oct 2025 00:53:42 GMT
• Title: RefusalBench: Generative Evaluation of Selective Refusal in Grounded Language Models
• Authors: Aashiq Muhamed, Leonardo F. R. Ribeiro, Markus Dreyer, Virginia Smith, Mona T. Diab,
• Abstract summary: The ability of language models to refuse to answer based on flawed systems remains a significant failure point.<n>We introduce RefusalBench, a generative methodology that creates diagnostic test cases through controlled linguistic context.<n>We find that selective refusal is a train, alignmentsensitive capability offering a clear path to improvement.
• Score: 43.76961935990733
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The ability of language models in RAG systems to selectively refuse to answer based on flawed context is critical for safety, yet remains a significant failure point. Our large-scale study reveals that even frontier models struggle in this setting, with refusal accuracy dropping below 50% on multi-document tasks, while exhibiting either dangerous overconfidence or overcaution. Static benchmarks fail to reliably evaluate this capability, as models exploit dataset-specific artifacts and memorize test instances. We introduce RefusalBench, a generative methodology that programmatically creates diagnostic test cases through controlled linguistic perturbation. Our framework employs 176 distinct perturbation strategies across six categories of informational uncertainty and three intensity levels. Evaluation of over 30 models uncovers systematic failure patterns: refusal comprises separable detection and categorization skills, and neither scale nor extended reasoning improves performance. We find that selective refusal is a trainable, alignment-sensitive capability, offering a clear path for improvement. We release two benchmarks -- RefusalBench-NQ (single document) and RefusalBench-GaRAGe (multi-document) -- and our complete generation framework to enable continued, dynamic evaluation of this critical capability.

Related papers

• Lost in the Noise: How Reasoning Models Fail with Contextual Distractors [57.31788955167306]
  Recent advances in reasoning models and agentic AI systems have led to an increased reliance on diverse external information.<n>We introduce NoisyBench, a comprehensive benchmark that systematically evaluates model robustness across 11 datasets in RAG, reasoning, alignment, and tool-use tasks.<n>Our evaluation reveals a catastrophic performance drop of up to 80% in state-of-the-art models when faced with contextual distractors.
  arXiv  Detail & Related papers  (2026-01-12T05:43:51Z)
• Rubric-Conditioned LLM Grading: Alignment, Uncertainty, and Robustness [4.129847064263056]
  We systematically evaluate the performance of Large Language Models for rubric-based short-answer grading.<n>We find that alignment is strong for binary tasks but degrades with increased rubric granularity.<n>Experiments reveal that while the model is resilient to prompt injection, it is sensitive to synonym substitutions.
  arXiv  Detail & Related papers  (2025-12-21T05:22:04Z)
• The NazoNazo Benchmark: A Cost-Effective and Extensible Test of Insight-Based Reasoning in LLMs [3.9977256267361754]
  We present Nazonazo, a cost-effective benchmark built from Japanese children's riddles to test insight-based reasoning.<n>No model except for GPT-5 is comparable to human performance, which achieves a 52.9% mean accuracy.
  arXiv  Detail & Related papers  (2025-09-18T07:50:04Z)
• ORFuzz: Fuzzing the "Other Side" of LLM Safety -- Testing Over-Refusal [27.26251627767238]
  Large Language Models (LLMs) increasingly exhibit over-refusal - erroneously rejecting benign queries due to overly conservative safety measures.<n>This paper introduces the first evolutionary testing framework, ORFuzz, for the systematic detection and analysis of LLM over-refusals.
  arXiv  Detail & Related papers  (2025-08-15T05:03:26Z)
• TrustLoRA: Low-Rank Adaptation for Failure Detection under Out-of-distribution Data [62.22804234013273]
  We propose a simple failure detection framework to unify and facilitate classification with rejection under both covariate and semantic shifts.<n>Our key insight is that by separating and consolidating failure-specific reliability knowledge with low-rank adapters, we can enhance the failure detection ability effectively and flexibly.
  arXiv  Detail & Related papers  (2025-04-20T09:20:55Z)
• On the Adversarial Robustness of Instruction-Tuned Large Language Models for Code [4.286327408435937]
  We assess the impact of diverse input challenges on the functionality and correctness of generated code using rigorous metrics and established benchmarks.<n>Open-source models demonstrate an increased susceptibility to input perturbations, resulting in declines in functional correctness ranging from 12% to 34%.<n>In contrast, commercial models demonstrate relatively greater resilience, with performance degradation ranging from 3% to 24%.
  arXiv  Detail & Related papers  (2024-11-29T07:00:47Z)
• Unsupervised Model Diagnosis [49.36194740479798]
  This paper proposes Unsupervised Model Diagnosis (UMO) to produce semantic counterfactual explanations without any user guidance. Our approach identifies and visualizes changes in semantics, and then matches these changes to attributes from wide-ranging text sources.
  arXiv  Detail & Related papers  (2024-10-08T17:59:03Z)
• Controlling Risk of Retrieval-augmented Generation: A Counterfactual Prompting Framework [77.45983464131977]
  We focus on how likely it is that a RAG model's prediction is incorrect, resulting in uncontrollable risks in real-world applications.<n>Our research identifies two critical latent factors affecting RAG's confidence in its predictions.<n>We develop a counterfactual prompting framework that induces the models to alter these factors and analyzes the effect on their answers.
  arXiv  Detail & Related papers  (2024-09-24T14:52:14Z)
• STOP! Benchmarking Large Language Models with Sensitivity Testing on Offensive Progressions [6.19084217044276]
  Mitigating explicit and implicit biases in Large Language Models (LLMs) has become a critical focus in the field of natural language processing.<n>We introduce the Sensitivity Testing on Offensive Progressions dataset, which includes 450 offensive progressions containing 2,700 unique sentences.<n>Our findings reveal that even the best-performing models detect bias inconsistently, with success rates ranging from 19.3% to 69.8%.
  arXiv  Detail & Related papers  (2024-09-20T18:34:38Z)
• Evaluating Language Model Context Windows: A "Working Memory" Test and Inference-time Correction [10.428174043080622]
  Large language models are prominently used in real-world applications, often tasked with reasoning over large volumes of documents. We propose SWiM, an evaluation framework that addresses the limitations of standard tests. We also propose medoid voting, a simple, but effective training-free approach that helps alleviate this effect.
  arXiv  Detail & Related papers  (2024-07-04T05:46:20Z)
• SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal [64.9938658716425]
  SORRY-Bench is a proposed benchmark for evaluating large language models' (LLMs) ability to recognize and reject unsafe user requests.<n>First, existing methods often use coarse-grained taxonomy of unsafe topics, and are over-representing some fine-grained topics.<n>Second, linguistic characteristics and formatting of prompts are often overlooked, like different languages, dialects, and more -- which are only implicitly considered in many evaluations.
  arXiv  Detail & Related papers  (2024-06-20T17:56:07Z)
• PAGER: A Framework for Failure Analysis of Deep Regression Models [27.80057763697904]
  We introduce PAGER (Principled Analysis of Generalization Errors in Regressors), a framework to systematically detect and characterize failures in deep regressors. Built upon the principle of anchored training in deep models, PAGER unifies both epistemic uncertainty and complementary manifold non-conformity scores to accurately organize samples into different risk regimes.
  arXiv  Detail & Related papers  (2023-09-20T00:37:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.