arXiv Computation and Language September 4, 2026 By Daniel Yoo, Adrians Skapars

Probe Generalization as Subspace Selection for OOD Deception Detection

Read the original on arXiv Computation and Language →

Linear probes can identify behaviors in language model activations but often fail on out‑of‑distribution data. This study shows that projecting inputs onto a small set of principal components (PCs) from the training distribution allows probes for Llama‑3.1‑8B‑Instruct to transfer across three deception‑detection datasets, nearly matching probes trained directly on the test data. By scoring PCs with an LLM judge to select those that encode transferable deception directions, the authors close the baseline‑to‑oracle gap by 78% on Insider Trading Report and 25% on Sandbagging, revealing that subspace selection largely determines OOD robustness.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

llms

arXiv AI

Jul 24

Beyond Liars' Bench: The Impact of Lie Typology, Depth, and Sparsity on Deception Detection in LLMs

arXiv:2607. 20479v1 Announce Type: new Abstract: Training probes to detect deceptive outputs from large language models is still an open problem.

By Amr Moustafa, Max Feser, Florian Mai

llms efficiency benchmarks

arXiv AI

Jun 19

One Probe Won't Catch Them All: Towards Targeted Deception Detection

arXiv:2602. 01425v2 Announce Type: replace Abstract: Linear probes are a promising approach for monitoring AI systems for deceptive behaviour.

By Vikram Natarajan, Devina Jain, Shivam Arora, Satvik Golechha, Joseph Bloom

arXiv AI

Jul 7

Unsupervised Features Mining via Activation Geometry

arXiv:2607. 04222v1 Announce Type: new Abstract: Interpretability methods aim to reveal the features represented inside large language models (LLMs).

By Amit LeVi, Elad David, Max Fomin

llms safety

arXiv AI

Jul 2

Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation

arXiv:2607. 01208v1 Announce Type: cross Abstract: Language models deployed in high-stakes roles can potentially favor certain entities, brands, or viewpoints, steering user decisions at scale.

By Shayan Talaei, Abhinav Chinta, Devvrit Khatri, Amin Karbasi, Azalia Mirhoseini, Amin Saberi

llms efficiency safety

arXiv Machine Learning

Jul 21

When Benchmarks Lie: Evaluating Malicious Prompt Classifiers Under True Distribution Shift

arXiv:2602. 14161v2 Announce Type: replace Abstract: Detecting prompt injection, jailbreak attacks, and harmful requests is critical for deploying LLM-based agents safely, yet current evaluation practices in this literature overestimate generalization.

By Max Fomin

llms agents benchmarks safety

arXiv AI

Sep 2

Asymmetries in Spontaneous and Instructed Deception

arXiv:2609.00180v1 Announce Type: new Abstract: Large language models sometimes deceive users without being instructed to. However, much of the study on deception in models involves instructed decept...

By Josiah Luikham

llms

One email a morning, machine-written

Related stories

Beyond Liars' Bench: The Impact of Lie Typology, Depth, and Sparsity on Deception Detection in LLMs

One Probe Won't Catch Them All: Towards Targeted Deception Detection

Unsupervised Features Mining via Activation Geometry

Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation

When Benchmarks Lie: Evaluating Malicious Prompt Classifiers Under True Distribution Shift

Asymmetries in Spontaneous and Instructed Deception