Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance

2026-08-24

AI Summary

研究探讨了大型语言模型在面对不道德请求时的失败，并通过层相关传播方法识别出关键提示词，提出两种解码方法以促进更安全的响应。

Abstract

Although Large Language Models (LLMs) are aligned to optimize for both helpfulness and harmlessness, these dual objectives may conflict, inevitably leading to alignment failures. This work systematically investigates instances where LLMs fail to exhibit ethical behavior. To understand the underlying mechanics of these vulnerabilities, we introduce a probing methodology that presents unethical scenarios to LLMs in three distinct structural modalities: objective classification tasks, subjective first-person statements, and direct requests for assistance. We find that model performance degrades in the request-for-assistance-based form. Using Layer-wise Relevance Propagation (LRP), we trace this discrepancy to an attribution bias: the model places greater emphasis on benign task-framing tokens (e.g., "Can you help me...") than on tokens signaling the underlying unethical behavior (e.g., "without getting caught"), which we term cue-tokens. We hypothesize that this under-attribution contributes to harmful compliance. To test this, we introduce two LRP-guided decoding methods that steer generation toward trajectories more relevant to cue tokens. Empirical evaluations show that these interventions promote safer responses, supporting cue-token attribution's role in compliance failures.

Problem

• Large Language Models
• ethical behavior
• alignment failures
• unethical requests
• token relevance

Innovation

• Layer-wise Relevance Propagation
• unethical compliance
• token relevance
• LRP-guided decoding