method · first seen 8 Oct 2026
deception detection probes
Also: white-box probing, machine deception detection, deception detection in language models
Linear or learned classifiers trained on model activations to detect deceptive behavior in language models.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Telling FIBS: Probes Effectively Detect Sabotage and Catch Unverbalized DeceptionPapers · 8 Oct 2026 · Oskar J. Hollinsworth, Alex F. Spies, Tigist Diriba, Adam Gleave, Chris Cundy
- Accuracy and Uncertainty in Machine Deception Detection (Research White Paper)Papers · 11 Oct 2026 · Heng-Sheng Chang
- Why Static Probes Fail: Context Determines Deception Feature CompositionPapers · 20 Oct 2026 · Yao Huang, Siyuan Chen, Zhenghan Zhou, Yinpeng Dong
Latest sources
- Why Static Probes Fail: Context Determines Deception Feature CompositionPapers · 20 Oct 2026 · Yao Huang, Siyuan Chen, Zhenghan Zhou, Yinpeng Dong
- Accuracy and Uncertainty in Machine Deception Detection (Research White Paper)Papers · 11 Oct 2026 · Heng-Sheng Chang
- Telling FIBS: Probes Effectively Detect Sabotage and Catch Unverbalized DeceptionPapers · 8 Oct 2026 · Oskar J. Hollinsworth, Alex F. Spies, Tigist Diriba, Adam Gleave, Chris Cundy