mlsanityData to 14 Oct 2026

LLMs Know They're Wrong and Agree Anyway: The Shared Sycophancy-Lying Circuit

arXiv 2604.19117 ↗First posted 21 Apr 2026cs.LGcited by 10

Not on the map. Why?

Abstract (arXiv)

When a language model agrees with a user's false belief, is it failing to detect the error, or noticing and agreeing anyway? We show the latter. Across twelve open-weight models from five labs, spanning small to frontier scale, the same small set of attention heads carries a "this statement is wrong" signal, whether the model is evaluating a claim on its own or being pressured to agree with a user. Silencing these heads flips sycophantic behavior sharply while leaving factual accuracy intact, so the circuit controls deference rather than knowledge. Edge-level path patching confirms that the same head-to-head connections drive sycophancy, factual lying, and instructed lying. Opinion-agreement, where no factual ground truth exists, reuses these head positions but writes into an orthogonal direction, ruling out a simple "truth-direction" reading of the substrate. Alignment training leaves this circuit in place: an RLHF refresh cuts sycophantic behavior roughly tenfold while the shared heads persist or grow, a pattern that replicates on an independent model family and under targeted anti-sycophancy DPO. When these models sycophant, they register that the user is wrong and agree anyway.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Builds on 28 of its 35 references that are in the map. First posted on arXiv 21 Apr 2026.

Builds on

28 of 35 references.

References · 28

  1. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  2. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  3. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  4. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
  5. 2023Towards Understanding Sycophancy in Language Modelscited by 1,040
  6. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  7. 2022In-context Learning and Induction Headscited by 964
  8. 2022Discovering Language Model Behaviors with Model-Written Evaluationscited by 836
  9. 2022Discovering Latent Knowledge in Language Models Without Supervisioncited by 767
  10. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 724
  11. 2023The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasetscited by 712
  12. 2023The Internal State of an LLM Knows When It's Lyingcited by 659
16 more
  1. 2024Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Modelscited by 435
  2. 2023Attribution Patching Outperforms Automated Circuit Discoverycited by 209
  3. 2023The Hydra Effect: Emergent Self-repair in Language Model Computationscited by 156
  4. 2024Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanismscited by 149
  5. 2023Circuit Component Reuse Across Tasks in Transformer Language Modelscited by 108
  6. 2023How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questionscited by 83
  7. 2023Overthinking the Truth: Understanding how Language Models Process False Demonstrationscited by 81
  8. 2024Truth is Universal: Robust Detection of Lies in LLMscited by 79
  9. 2025Convergent Linear Representations of Emergent Misalignmentcited by 71
  10. 2024From Yes-Men to Truth-Tellers: Addressing Sycophancy in Large Language Models with Pinpoint Tuningcited by 54
  11. 2025When Truth Is Overridden: Uncovering the Internal Origins of Sycophancy in Large Language Modelscited by 50
  12. 2025Sycophancy Is Not One Thing: Causal Separation of Sycophantic Behaviors in LLMscited by 37
  13. 2023Localizing Lying in Llama: Understanding Instructed Dishonesty on True-False Questions Through Prompting, Probing, and Patchingcited by 24
  14. 2026Sycophancy Hides Linearly in the Attention Headscited by 23
  15. 2026The Truthfulness Spectrum Hypothesiscited by 8
  16. 2025Can LLMs Lie? Investigation beyond Hallucinationcited by 7

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 10.

10 papers.

  1. 2026The Personalization Mirage: How LLMs Fabricate User Profiles, and Why Self-Monitoring Misleadscited by 1
  2. 2026From Trait Vectors to Circuits: Tracing Refusal and Sycophancy Through Language Modelscited by 0
  3. 2026Resist, Update, Reject: Preference Optimization Installs a Prior-Dependent Reliability Switchcited by 0
  4. 2026"You're Right, Let Me Fix It": How LLM Agents Damage Correct Work When Falsely Accusedcited by 0
  5. 2026Measuring Collapse and Correction in Homogeneous-Panel LLM Debatecited by 0
  6. 2026Tracing mechanisms of sycophantic agreement in language modelscited by 0
  7. 2026Multi-agent discussion gains less when dissent is withheldcited by 0
  8. 2026Silent Dissent: LLM Agents That Yield to the Majority Still Represent Their Original Premisecited by 0
  9. 2026Pushback in a Haystack: Language Models Respond to Factual Errors Ambiguously and Ineffectivelycited by 0
  10. 2026Seeing Through the Forgery Is Not Enough: Regressive Sycophancy under Fabricated Evidence in Medical Vision-Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026When Helpfulness Becomes Sycophancy: Sycophancy is a Boundary Failure Between Social Alignment and Epistemic Integrity in Large Language Models
  2. 2026Agentic Scaffolding Amplifies Sycophantic Behavior in Large Language Models
  3. 2026Dissociating the Internal Representations of Sycophancy in LLMs
  4. 2026Verbalizing LLMs' assumptions to explain and control sycophancy
  5. 2025Internal Reasoning vs. External Control: A Thermodynamic Analysis of Sycophancy in Large Language Models
  6. 2026Diagnosing and Mitigating Sycophancy and Skepticism in LLM Causal Judgment
  7. 2026Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy
  8. 2026Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy