Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges

• URL: http://arxiv.org/abs/2605.26156v2
• Date: Thu, 28 May 2026 09:12:16 GMT
• Title: Turning Bias into Bugs: Bandit-Guided Style Manipulation Attacks on LLM Judges
• Abstract summary: We introduce BITE, a black-box adversarial framework that learns semantics-preserving edits to mislead an LLM judge.<n>BITE achieves an attack success rate exceeding 65% and raises scores by 1-2 points on a 9-point scale.<n>Our findings expose a fundamental weakness in the LLM-as-a-judge paradigm and motivate robust, attack-aware evaluation.
• Score: 65.92623263645139
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The known stylistic biases in LLM judges, such as a preference for verbosity or specific sentence structures, present an underexplored security vulnerability. In this work, we introduce BITE (BIas exploraTion and Exploitation), a black-box adversarial framework that learns semantics-preserving edits to mislead an LLM judge and artificially inflate the scores it assigns. We cast the selection of stylistic edits as a contextual bandit problem and use a LinUCB policy to adaptively choose edits that maximize the judge's score without access to model parameters or gradients. Empirically, we test BITE across a diverse range of LLM judges and tasks, including both pointwise and pairwise comparisons on chatbot leaderboards and AI-reviewer benchmarks. BITE achieves an attack success rate exceeding 65% and raises scores by 1-2 points on a 9-point scale, all while preserving semantic equivalence. We further assess the attack's stealthiness, showing that BITE evades standard style-control methods and several detection baselines. Our findings expose a fundamental weakness in the LLM-as-a-judge paradigm and motivate robust, attack-aware evaluation. Our code is available at https://github.com/xianglinyang/llm-as-a-judge-attack.

Related papers

• LivingArena: Do LLMs Know What Other LLMs Don't? Peer-Probing as Scalable Evaluation [79.03892269184145]
  LivingArena is an automated, contamination-resistant evaluation framework.<n>Models take turns proposing questions, aiming to pose items that opponents cannot answer correctly.<n>Questioners are encouraged to actively identify and exploit opponents' knowledge boundaries.
  arXiv  Detail & Related papers  (2026-06-19T06:20:58Z)
• A Coin Flip for Safety: LLM Judges Fail to Reliably Measure Adversarial Robustness [57.510025257780306]
  We show that existing validation protocols fail to account for substantial distribution shifts inherent to red-teaming.<n>We propose ReliableBench, a benchmark of behaviors that remain more consistently judgeable, and JudgeStressTest, a dataset designed to expose judge failures.
  arXiv  Detail & Related papers  (2026-02-04T15:13:35Z)
• Fooling LLM graders into giving better grades through neural activity guided adversarial prompting [26.164839501935973]
  We propose a systematic method to reveal such biases in AI evaluation systems.<n>Our approach first identifies hidden neural activity patterns that predict distorted decision outcomes.<n>We demonstrate that this combination can effectively fool large language model graders into assigning much higher grades than humans would.
  arXiv  Detail & Related papers  (2024-12-17T19:08:22Z)
• Style Outweighs Substance: Failure Modes of LLM Judges in Alignment Benchmarking [56.275521022148794]
  Post-training methods claim superior alignment by virtue of better correspondence with human pairwise preferences.<n>We attempt to answer the question -- do LLM-judge preferences translate to progress on other, more concrete metrics for alignment, and if not, why not?<n>We find that (1) LLM-judge preferences do not correlate with concrete measures of safety, world knowledge, and instruction following; (2) LLM-judges have powerful implicit biases, prioritizing style over factuality and safety; and (3) the supervised fine-tuning stage of post-training, and not the PO stage, has the greatest impact on alignment.
  arXiv  Detail & Related papers  (2024-09-23T17:58:07Z)
• Evaluating Implicit Bias in Large Language Models by Attacking From a Psychometric Perspective [66.34066553400108]
  We conduct a rigorous evaluation of large language models' implicit bias towards certain demographics.<n>Inspired by psychometric principles, we propose three attack approaches, i.e., Disguise, Deception, and Teaching.<n>Our methods can elicit LLMs' inner bias more effectively than competitive baselines.
  arXiv  Detail & Related papers  (2024-06-20T06:42:08Z)
• BruSLeAttack: A Query-Efficient Score-Based Black-Box Sparse Adversarial Attack [22.408968332454062]
  We study the unique, less-well understood problem of generating sparse adversarial samples simply by observing the score-based replies to model queries. We develop the BruSLeAttack-a new, faster (more query-efficient) algorithm for the problem. Our work facilitates faster evaluation of model vulnerabilities and raises our vigilance on the safety, security and reliability of deployed systems.
  arXiv  Detail & Related papers  (2024-04-08T08:59:26Z)
• Is LLM-as-a-Judge Robust? Investigating Universal Adversarial Attacks on Zero-shot LLM Assessment [8.948475969696075]
  Large Language Models (LLMs) are powerful zero-shot assessors used in real-world situations such as assessing written exams and benchmarking systems. We show that short universal adversarial phrases can be deceived to judge LLMs to predict inflated scores. It is found that judge-LLMs are significantly more susceptible to these adversarial attacks when used for absolute scoring.
  arXiv  Detail & Related papers  (2024-02-21T18:55:20Z)
• Zero-Query Transfer Attacks on Context-Aware Object Detectors [95.18656036716972]
  Adversarial attacks perturb images such that a deep neural network produces incorrect classification results. A promising approach to defend against adversarial attacks on natural multi-object scenes is to impose a context-consistency check. We present the first approach for generating context-consistent adversarial attacks that can evade the context-consistency check.
  arXiv  Detail & Related papers  (2022-03-29T04:33:06Z)

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.