Article

How LLMs Audit Each Other: Five Mechanisms of Auditor Bias in Cross-Model Peer Review Under Identity Disclosure and Cross-Lingual Conditions

2026-06-30

Abstract excerpt

<title>Abstract</title> <p>As large language models (LLMs) are increasingly proposed for automated evaluation of other LLMs, a foundational question remains unanswered: do these models audit consistently when they know the identity of the system they are evaluating, and when the language of the auditing instruction does not match the language of the audited evidence? This paper addresses both questions through th...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: 4e85d8e9-d815-5fe4-ac92-5214566f6ef8
DOI: 10.21203/rs.3.rs-10171795/v1

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

How LLMs Audit Each Other: Five Mechanisms of Auditor Bias in Cross-Model Peer Review Under Identity Disclosure and Cross-Lingual ConditionsDOI 10.21203/rs.3.rs-10171795/v1Read

Verifier Exploitation in NLI-Guided Iterative Refinement: A Controlled Empirical AnalysisSemantic proximity Read

From Output Errors to Workflow Harm: A Practitioner-Audit Method for LLM-Mediated ResearchSemantic proximity Read

Quantifying and Mitigating Gender Bias in Legal Large Language Models:A Counterfactual Fairness FrameworkSemantic proximity Read

Explainable Automatic Evaluation: Interpretable Decision-Making in Large Language Model AutoratersSemantic proximity Read

Limits of Self-Correction in LLMs: An Information-Theoretic Analysis of Correlated ErrorsSemantic proximity Read

From Linguistic Rights to Explainable Learning Analytics: A Cross-jurisdictional Policy Framework for Generative AI Feedback in Professional Interpreter EducationSemantic proximity Read

The Dice Roll Method: A StandardizedProtocol for Measuring Stochastic Bias in Large Language Model OutputSemantic proximity Read

RECON: A Recipe-Driven, Evidence-Fused, Neuro-Symbolic Multi-Agent Architecture for Autonomous Document Compliance Auditing across Heterogeneous Enterprise RepositoriesSemantic proximity Read

AiDER: Auditing and Document Evaluation via Rule Compilation and Small Language Models - An Education Case StudySemantic proximity Read

Harm-Conditioned Computational Friction as a Contrastive Diagnostic of Safety Representations in Language Models: A Proof-of-Concept and Control AuditSemantic proximity Read

Classifying 25 Misinterpretations of Statistical Tests: A Comparison of Six Large Language ModelsSemantic proximity Read

Graduated Dissent: Budgeted Disagreement Resolution for Multi-Model InferenceSemantic proximity Read