mlsanityData to 7 Oct 2026

Disentangling Deception and Hallucination Failures in LLMs

arXiv 2602.14529 ↗First posted 16 Feb 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Failures in large language models (LLMs) are often analyzed from a behavioral perspective, where incorrect outputs in factual question answering are commonly associated with missing knowledge. In this work, focusing on entity-based factual queries, we suggest that such a view may conflate different failure mechanisms, and propose an internal, mechanism-oriented perspective that separates Knowledge Existence from Behavior Expression. Under this formulation, hallucination and deception correspond to two qualitatively different failure modes that may appear similar at the output level but differ in their underlying mechanisms. To study this distinction, we construct a controlled environment for entity-centric factual questions in which knowledge is preserved while behavioral expression is selectively altered, enabling systematic analysis of four behavioral cases. We analyze these failure modes through representation separability, sparse interpretability, and inference-time activation steering.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 80 of its 91 references that are in the map. First posted on arXiv 16 Feb 2026.

Builds on

80 of 91 references.

References · 80

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  4. 2025Qwen3 Technical Reportcited by 8,195
  5. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  6. 2017Deep reinforcement learning from human preferencescited by 3,846
  7. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  8. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  9. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,359
  10. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
  11. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,081
  12. 2023Towards Understanding Sycophancy in Language Modelscited by 988
38 more
  1. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
  2. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
  3. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 948
  4. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 946
  5. 2023Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formattingcited by 778
  6. 2023Jailbreaking ChatGPT via Prompt Engineering: An Empirical Studycited by 478
  7. 2023GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Ciphercited by 401
  8. 2023The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"cited by 399
  9. 2023Understanding the Effects of RLHF on LLM Generalisation and Diversitycited by 386
  10. 2023Towards Best Practices of Activation Patching in Language Models: Metrics and Methodscited by 328
  11. 2023Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoningcited by 326
  12. 2024INSIDE: LLMs' Internal States Retain the Power of Hallucination Detectioncited by 317
  13. 2024Alignment faking in large language modelscited by 316
  14. 2023How Language Model Hallucinations Can Snowballcited by 310
  15. 2024Frontier Models are Capable of In-context Schemingcited by 252
  16. 2024Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMscited by 246
  17. 2024LLMs Know More Than They Show: On the Intrinsic Representation of LLM Hallucinationscited by 230
  18. 2024OR-Bench: An Over-Refusal Benchmark for Large Language Modelscited by 222
  19. 2024Jailbreak Attacks and Defenses Against Large Language Models: A Surveycited by 215
  20. 2024Programming Refusal with Conditional Activation Steeringcited by 191
  21. 2024Improving Instruction-Following in Language Models through Activation Steeringcited by 149
  22. 2023Shifting Attention to Relevance: Towards the Predictive Uncertainty Quantification of Free-Form Large Language Modelscited by 147
  23. 2024WARM: On the Benefits of Weight Averaged Reward Modelscited by 145
  24. 2024FinBen: A Holistic Financial Benchmark for Large Language Modelscited by 140
  25. 2023Large Language Models can Strategically Deceive their Users when Put Under Pressurecited by 132
  26. 2024AI Sandbagging: Language Models can Strategically Underperform on Evaluationscited by 131
  27. 2023Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigationcited by 123
  28. 2024Language Models Learn to Mislead Humans via RLHFcited by 118
  29. 2024A Survey on Uncertainty Quantification of Large Language Models: Taxonomy, Open Research Challenges, and Future Directionscited by 115
  30. 2024Do I Know This Entity? Knowledge Awareness and Hallucinations in Language Modelscited by 114
  31. 2024Secret Collusion among AI Agents: Multi-Agent Deception via Steganographycited by 107
  32. 2025The Geometry of Refusal in Large Language Models: Concept Cones and Representational Independencecited by 104
  33. 2024Improved Techniques for Optimization-Based Jailbreaking on Large Language Modelscited by 98
  34. 2024The Dawn After the Dark: An Empirical Study on Factuality Hallucination in Large Language Modelscited by 87
  35. 2024Truth is Universal: Robust Detection of Lies in LLMscited by 78
  36. 2023On Measuring Faithfulness or Self-consistency of Natural Language Explanationscited by 71
  37. 2024Feedback Loops With Language Models Drive In-Context Reward Hackingcited by 70
  38. 2024SuRe: Summarizing Retrievals using Answer Candidates for Open-domain QA of LLMscited by 67

50 most cited of 80 All 80 (JSON)

Built on by

Not cited yet.