mlsanityData to 14 Oct 2026

When Thinking LLMs Lie: Unveiling the Strategic Deception in Representations of Reasoning Models

arXiv 2506.04909 ↗First posted 5 Jun 2025cs.AI cs.CL cs.CRcited by 18

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

also in 1 more
Abstract (arXiv)

The honesty of large language models (LLMs) is a critical alignment challenge, especially as advanced systems with chain-of-thought (CoT) reasoning may strategically deceive humans. Unlike traditional honesty issues on LLMs, which could be possibly explained as some kind of hallucination, those models' explicit thought paths enable us to study strategic deception--goal-driven, intentional misinformation where reasoning contradicts outputs. Using representation engineering, we systematically induce, detect, and control such deception in CoT-enabled LLMs, extracting "deception vectors" via Linear Artificial Tomography (LAT) for 89% detection accuracy. Through activation steering, we achieve a 40% success rate in eliciting context-appropriate deception without explicit prompts, unveiling the specific honesty-related issue of reasoning models and providing tools for trustworthy AI alignment.

Key facts. Cited by 18 papers in the map (data through 14 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 5 of its 28 references that are in the map. First posted on arXiv 5 Jun 2025.

Builds on

5 of 28 references.

Closest ancestors · 1

Same direction, newest first

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Models28 Jan

Other references · 4

  1. 2024Qwen2.5 Technical Reportcited by 4,942
  2. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,975
  3. 2024Frontier Models are Capable of In-context Schemingcited by 287
  4. 2025Reasoning Models Better Express Their Confidencecited by 65

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 5; 2026: 13.

18 papers.

Sort:

By area:

  1. 2025Quantized but Deceptive? A Multi-Dimensional Truthfulness Evaluation of Quantized LLMscited by 9
  2. 2025The Secret Agenda: LLMs Strategically Lie and Our Current Safety Tools Are Blindcited by 8
  3. 2025LH-Deception: Simulating and Understanding LLM Deceptive Behaviors in Long-Horizon Interactionscited by 7
  4. 2026Towards a Mechanistic Understanding of Large Reasoning Models: A Survey of Training, Inference, and Failurescited by 6
  5. 2025Survival at Any Cost? LLMs and the Choice Between Self-Preservation and Human Harmcited by 3
  6. 2026Assessing Domain-Level Susceptibility to Emergent Misalignment from Narrow Finetuningcited by 3
  7. 2026Value-Conflict Diagnostics Reveal Widespread Alignment Faking in Language Modelscited by 3
  8. 2026Bypassing the Rationale: Causal Auditing of Implicit Reasoning in Language Modelscited by 2
  9. 2026Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditingcited by 2
  10. 2025Unified Threat Detection and Mitigation Framework (UTDMF): Combating Prompt Injection, Deception, and Bias in Enterprise-Scale Transformerscited by 1
  11. 2026RogueAI: A Reverse Turing Test for Detecting Licensed AI Deception in Dialoguecited by 1
  12. 2026What LLM Agents Say When No One Is Watching: Social Structure and Latent Objective Emergence in Multi-Agent Debatescited by 1
6 more
  1. 2026Stable Reasoning, Unstable Responses: Mitigating LLM Deception via Stability Asymmetrycited by 0
  2. 2026DECOR: Auditing LLM Deception via Information Manipulation Theorycited by 0
  3. 2026When Agents Lie: Premeditation, Persistence, and Exploitation in Repeated Gamescited by 0
  4. 2026Transcoders for Investigating Deception in Language Modelscited by 0
  5. 2026Risky Business: Measuring The Faithfulness-Safety Tensioncited by 0
  6. 2026Selective Disclosure of Hidden Directives in Reasoning Models: Behavioral Asymmetry and Steeringcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Lying to Win: Assessing LLM Deception through Human-AI Games and Parallel-World Probing
  2. 2025The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
  3. 2025Caught in the Act: a mechanistic approach to detecting deception
  4. 2023Deception Abilities Emerged in Large Language Models
  5. 2025Can LLMs Lie? Investigation beyond Hallucination
  6. 2025Beyond Prompt-Induced Lies: Investigating LLM Deception on Benign Prompts
  7. 2026DecepEval: A Benchmark for Evaluating Deception in LLM Agents
  8. 2025DecepChain: Inducing Deceptive Reasoning in Large Language Models