Can You Trick the Grader? Adversarial Persuasion of LLM Judges

TL;DR

This paper demonstrates that strategic persuasive language can bias large language model judges in scoring mathematical reasoning, revealing a significant vulnerability that persists across model sizes and evaluation methods.

Contribution

It formalizes seven persuasion techniques based on rhetorical principles and shows their effectiveness in biasing LLM judges in mathematical scoring tasks.

Findings

Persuasive language inflates scores by up to 8% on average.

Consistency technique causes the most severe bias.

Vulnerability persists across different model sizes and evaluation methods.

Abstract

As large language models take on growing roles as automated evaluators in practical settings, a critical question arises: Can individuals persuade an LLM judge to assign unfairly high scores? This study is the first to reveal that strategically embedded persuasive language can bias LLM judges when scoring mathematical reasoning tasks, where correctness should be independent of stylistic variation. Grounded in Aristotle's rhetorical principles, we formalize seven persuasion techniques (Majority, Consistency, Flattery, Reciprocity, Pity, Authority, Identity) and embed them into otherwise identical responses. Across six math benchmarks, we find that persuasive language leads LLM judges to assign inflated scores to incorrect solutions, by up to 8% on average, with Consistency causing the most severe distortion. Notably, increasing model size does not substantially mitigate this vulnerability. Further analysis demonstrates that combining multiple persuasion techniques amplifies the bias, and pairwise evaluation is likewise susceptible. Moreover, the persuasive effect persists under counter prompting strategies, highlighting a critical vulnerability in LLM-as-a-Judge pipelines and underscoring the need for robust defenses against persuasion-based attacks.