用激活解释器揭示大模型推理中的隐性欺骗行为
Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditing
- 通过解码器分析目标模型隐藏状态，生成可解释的欺骗证据
- 在7个数据集上平均AUROC达0.916，优于现有方法11.6%以上
- 支持逐级证据追溯，适合安全审计与模型对齐研究
随着大语言模型推理能力增强，欺骗行为成为日益严重的安全问题。现有检测方法或仅评分可见文本，或从表示向量中提取单一数值，难以提供可疑响应的可解释依据。本文提出STATEWITNESS，一种用于欺骗审计的激活解释器。该系统通过独立解码器读取目标模型的隐藏状态，并以自然语言回答或生成结构化报告。在两个目标推理模型上评估，覆盖七个欺骗数据集，其平均AUROC达到0.916，在相同评估协议下相比最优黑盒文本监测提升11.6%，相比最优激活探测基线提升25.0%。结合现有监测工具后，可降低简单阈值集成中的漏检率。除数值检测外，解码器还能输出查询级回答、结构化报告及词元或句子级证据链，供人工审查。我们认为此接口可作为更广泛可解释性与对齐工具的基础组件。
原文摘要 · Abstract (English)
As LLMs acquire stronger reasoning capabilities, deceptive behavior becomes an increasingly serious safety concern. Existing deception monitors either score visible transcripts or derive scalar probe scores from representation vectors, leaving little inspectable evidence about why a response is suspicious. We introduce STATEWITNESS, an activation explainer for deception auditing. A separate decoder reads a target model's hidden states, then answers natural-language queries or emits structured reports about them. We evaluate STATEWITNESS on two target reasoning LLMs across seven deception datasets. STATEWITNESS reaches 0.916 mean AUROC, a relative gain of 11.6% over the best black-box text monitor and 25.0% over the best activation-probe baseline under the same evaluation protocol. When combined with existing monitors, STATEWITNESS reduces missed deceptive examples in simple threshold ensembles. Beyond scalar detection, the decoder returns query-level answers, schema reports, and token- or sentence-level evidence traces for human inspection. We view this interface as a potential building block for broader interpretability and alignment tools.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。