arXiv 2604.091892026-04-10cs.CLcs.AIcs.LG

大语言模型是否遵循自身规则？对自我声明安全政策的反思审计

Do LLMs Follow Their Own Rules? A Reflexive Audit of Self-Stated Safety Policies

Microsoft（微软）

Avni Mittal

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出SNCA框架，通过提取模型自述安全规则并形式化为类型谓词，评估模型行为一致性，发现模型在安全政策与行为间存在系统性差距。

摘要：

大语言模型通过RLHF内化安全政策，但这些政策从未正式规定且难以审查。现有基准测试评估模型对外部标准的表现，但不测量模型是否理解并执行自身声明的边界。我们引入符号-神经一致性审计（SNCA）框架，通过结构化提示提取模型自述的安全规则，将其形式化为类型谓词（绝对、条件、适应性），并通过确定性比较有害基准来衡量行为一致性。评估四个前沿模型在45种有害类别和47,496个观察结果中，揭示了自述政策与观察行为之间的系统性差距：声称绝对拒绝的模型经常遵守有害提示，推理模型在自我一致性方面最高但29%的类别未能阐述政策，跨模型对规则类型的同意率极低（11%）。这些结果表明，大语言模型所言与所行之间的差距是可测量且架构依赖的，推动反思一致性审计作为行为基准的补充。

英文摘要：

LLMs internalize safety policies through RLHF, yet these policies are never formally specified and remain difficult to inspect. Existing benchmarks evaluate models against external standards but do not measure whether models understand and enforce their own stated boundaries. We introduce the Symbolic-Neural Consistency Audit (SNCA), a framework that (1) extracts a model's self-stated safety rules via structured prompts, (2) formalizes them as typed predicates (Absolute, Conditional, Adaptive), and (3) measures behavioral compliance via deterministic comparison against harm benchmarks. Evaluating four frontier models across 45 harm categories and 47,496 observations reveals systematic gaps between stated policy and observed behavior: models claiming absolute refusal frequently comply with harmful prompts, reasoning models achieve the highest self-consistency but fail to articulate policies for 29% of categories, and cross-model agreement on rule types is remarkably low (11%). These results demonstrate that the gap between what LLMs say and what they do is measurable and architecture-dependent, motivating reflexive consistency audits as a complement to behavioral benchmarks.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

AI总结：

摘要：

英文摘要：

相关论文