mlsanityData to 14 Oct 2026

Localizing Lying in Llama: Understanding Instructed Dishonesty on True-False Questions Through Prompting, Probing, and Patching

arXiv 2311.15131 ↗First posted 25 Nov 2023cs.LG cs.AI cs.CLcited by 24

On the map: Mechanistic interpretability → Probing deception, bias and beliefs

Abstract (arXiv)

Large language models (LLMs) demonstrate significant knowledge through their outputs, though it is often unclear whether false outputs are due to a lack of knowledge or dishonesty. In this paper, we investigate instructed dishonesty, wherein we explicitly prompt LLaMA-2-70b-chat to lie. We perform prompt engineering to find which prompts best induce lying behavior, and then use mechanistic interpretability approaches to localize where in the network this behavior occurs. Using linear probing and activation patching, we localize five layers that appear especially important for lying. We then find just 46 attention heads within these layers that enable us to causally intervene such that the lying model instead answers honestly. We show that these interventions work robustly across many prompts and dataset splits. Overall, our work contributes a greater understanding of dishonesty in LLMs so that we may hope to prevent it.

Key facts. Cited by 24 papers in the map (data through 14 Oct 2026). Placed in Probing deception, bias and beliefs, one of 13 directions in Mechanistic interpretability. Builds on 6 of its 13 references that are in the map. First posted on arXiv 25 Nov 2023.

Builds on

6 of 13 references.

References · 6

  1. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  2. 2023The Internal State of an LLM Knows When It's Lyingcited by 659
  3. 2023An Overview of Catastrophic AI Riskscited by 222
  4. 2023How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questionscited by 83
  5. 2023Overthinking the Truth: Understanding how Language Models Process False Demonstrationscited by 81
  6. 2023Still No Lie Detector for Language Models: Probing Empirical and Conceptual Roadblockscited by 78

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 11; 2025: 9; 2026: 4.

24 papers.

Sort:

By area:

  1. 2024Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modificationscited by 231
  2. 2024On the Role of Attention Heads in Large Language Model Safetycited by 72
  3. 2025The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systemscited by 72
  4. 2024Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Eracited by 46
  5. 2024Model Editing as a Robust and Denoised variant of DPO: A Case Study on Toxicitycited by 35
  6. 2024Standards for Belief Representations in LLMscited by 30
  7. 2024AI Risk Management Should Incorporate Both Safety and Securitycited by 17
  8. 2025Difficulties with Evaluating a Deception Detector for AIscited by 13
  9. 2024LLMs Faithfully and Iteratively Compute Answers During CoT: A Systematic Analysis With Multi-step Arithmeticscited by 11
  10. 2026LLMs Know They're Wrong and Agree Anyway: The Shared Sycophancy-Lying Circuitcited by 10
  11. 2025Predicting the Performance of Black-box LLMs through Follow-up Queriescited by 9
  12. 2025Improving Preference Extraction In LLMs By Identifying Latent Knowledge Through Classifying Probescited by 9
12 more
  1. 2025Exploring and Exploiting the Inherent Efficiency within Large Reasoning Models for Self-Guided Efficiency Enhancementcited by 8
  2. 2024Large Language Models as Misleading Assistants in Conversationcited by 7
  3. 2025Can LLMs Lie? Investigation beyond Hallucinationcited by 7
  4. 2025Rep2Text: Decoding Full Text from a Single LLM Token Representationcited by 7
  5. 2024Lens: Rethinking Multilingual Enhancement for Large Language Modelscited by 6
  6. 2024Cluster-norm for Unsupervised Probing of Knowledgecited by 4
  7. 2025This Is Your Doge, If It Please You: Exploring Deception and Robustness in Mixture of LLMscited by 4
  8. 2024Logistic Regression makes small LLMs strong and explainable "tens-of-shot" classifierscited by 3
  9. 2025Trick or Neat: Adversarial Ambiguity and Language Model Evaluationcited by 3
  10. 2026Detecting Contextual Hallucinations in LLMs with Frequency-Aware Attentioncited by 2
  11. 2026Nodes Are Early, Edges Are Late: Probing Diagram Representations in Large Vision-Language Modelscited by 1
  12. 2026A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Revealcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Beyond Prompt-Induced Lies: Investigating LLM Deception on Benign Prompts
  2. 2026Lying to Win: Assessing LLM Deception through Human-AI Games and Parallel-World Probing
  3. 2025When Thinking LLMs Lie: Unveiling the Strategic Deception in Representations of Reasoning Models
  4. 2026DECOR: Auditing LLM Deception via Information Manipulation Theory
  5. 2026"Did you lie?" Evaluating Lie Detectors across Model Scale and Belief-Verified Model Organisms
  6. 2026Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditing
  7. 2026Disentangling Deception and Hallucination Failures in LLMs
  8. 2026Asymmetries in Spontaneous and Instructed Deception