← 研踪 PaperTrailACL Anthology · ACL 2026:acl:2026.acl-long.1582 · 记录 #1
Trajectory Signatures of Deception in Large Language Models
来源日期（仅到月份） 2026-07
来源记录日期 2026-07
雷达观测 2026-11-01 21:52
摘要
Detecting deceptive behavior in LLMs is typically done post-hoc on outputs or by probing static activations. We instead treat deception as a dynamic process, a trajectory through the model’s hidden-state space during inference. We capture layerwise activations at sparse “decision points” where the model is uncertain between competing tokens, forming activation trajectories for matched truthful vs. deceptive responses across strategic deception, sycophancy, instructed deception, and confabulation. Across GPT-2 and Llama variants, deceptive generation is associated with changes in trajectory geometry, but increases in path length are model and deception-type-dependent. Sycophancy shows the clearest signal, whereas instructed deception yields near-null signatures. With just 7 geometric features, a lightweight classifier achieves performance comparable to PCA-reduced probing at matched dimensionality for binary sycophancy detection and shows preliminary utility for 4-way deception-type classification. These findings indicate that trajectory-based monitoring can provide process-level signals associated with deceptive generation during inference, complementing methods that focus on endpoint activation states.
论文发现出处
这些页面提供论文线索，不代表同行评审或质量认证。
观测到的版本
平台记录号代表本站观测到的元数据变化，不是作者修订号。来源日期仅到月份或年份时，按该月或该年首日存档。
记录 #1 · 2026-07 · 最新观测版本
Trajectory Signatures of Deception in Large Language Models
Detecting deceptive behavior in LLMs is typically done post-hoc on outputs or by probing static activations. We instead treat deception as a dynamic process, a trajectory through the model’s hidden-state space during inference. We capture layerwise activations at sparse “decision points” where the model is uncertain between competing tokens, forming activation trajectories for matched truthful vs. deceptive responses across strategic deception, sycophancy, instructed deception, and confabulation. Across GPT-2 and Llama variants, deceptive generation is associated with changes in trajectory geometry, but increases in path length are model and deception-type-dependent. Sycophancy shows the clearest signal, whereas instructed deception yields near-null signatures. With just 7 geometric features, a lightweight classifier achieves performance comparable to PCA-reduced probing at matched dimensionality for binary sycophancy detection and shows preliminary utility for 4-way deception-type classification. These findings indicate that trajectory-based monitoring can provide process-level signals associated with deceptive generation during inference, complementing methods that focus on endpoint activation states.
查看此版本解读 →查看此版本原文 ↗