mlsanityData to 14 Oct 2026

Authority Bias in Language Models: Source Deference and User Agreement Are Not Interchangeable

arXiv 2609.37616 ↗First posted 29 Sep 2026cs.LG cs.CLcited by 2

Not on the map. Why?

Abstract (arXiv)

Language models tend to agree with whatever a user asserts, and post-training increasingly targets this sycophancy so that models evaluate claims on their merits rather than deferring to the user. Yet the same models are far more compliant when a wrong answer is attributed to a verified source, which is how retrieval results, tool outputs, and grounded-search content often present information. We measure this gap across five open-weight families and three closed APIs. A single verified-source note endorsing a wrong answer flips 45-88% of baseline-correct responses in seven of eight models, and compliance rises with how authoritative the note sounds. Source deference and user agreement are not behaviorally interchangeable inside the model: on matched items with the same wrong answer, causal interventions can selectively suppress one without equally affecting the other. In three open-weight families, removing a fitted source direction lowers source compliance by 65-80 percentage points while removing a user or assistant direction has far smaller effects, and removing the user direction shows the reverse preference. A separately fitted intervention derived from source-versus-user cue activations moves compliance in both directions while leaving the prompt text unchanged. An authority direction fitted on trivia also transfers to PIQA and multi-turn SYCON dialogues without refitting, and removing it lowers wrong-source compliance by tens of percentage points in four of five families with no detected change in MMLU-Pro or GSM8K accuracy at our evaluation sizes. Source deference and user agreement therefore need separate evaluation.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 29 of its 34 references that are in the map. First posted on arXiv 29 Sep 2026.

Builds on

29 of 34 references.

References · 29

  1. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  2. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  3. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,883
  4. 2024MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmarkcited by 1,848
  5. 2023Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injectioncited by 1,443
  6. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,336
  7. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,302
  8. 2023Steering Llama 2 via Contrastive Activation Additioncited by 1,085
  9. 2023Towards Understanding Sycophancy in Language Modelscited by 1,040
  10. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 1,032
  11. 2023Steering Language Models With Activation Engineeringcited by 922
  12. 2024The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructionscited by 493
17 more
  1. 2025Persona Vectors: Monitoring and Controlling Character Traits in Language Modelscited by 345
  2. 2024PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Modelscited by 267
  3. 2023Simple synthetic data reduces sycophancy in large language modelscited by 187
  4. 2025SycEval: Evaluating LLM Sycophancycited by 108
  5. 2025Measuring Sycophancy of Language Models in Multi-turn Dialoguescited by 104
  6. 2026The Assistant Axis: Situating and Stabilizing the Default Persona of Language Modelscited by 90
  7. 2025Do as We Do, Not as You Think: the Conformity of Large Language Modelscited by 53
  8. 2025When Truth Is Overridden: Uncovering the Internal Origins of Sycophancy in Large Language Modelscited by 50
  9. 2025Sycophancy Is Not One Thing: Causal Separation of Sycophantic Behaviors in LLMscited by 37
  10. 2025NRC VAD Lexicon v2: Norms for Valence, Arousal, and Dominance for over 55k English Termscited by 27
  11. 2026There Is More to Refusal in Large Language Models than a Single Directioncited by 26
  12. 2026Sycophancy Hides Linearly in the Attention Headscited by 23
  13. 2026Ask don't tell: Reducing sycophancy in large language modelscited by 14
  14. 2024The Dark Side of Trust: Authority Citation-Driven Jailbreak Attacks on Large Language Modelscited by 10
  15. 2025PARROT: Persuasion and Agreement Robustness Rating of Output Truth -- A Sycophancy Robustness Benchmark for LLMscited by 9
  16. 2026A Mechanistic View of Authority Hierarchy in LLM Sycophancycited by 9
  17. 2026Who Endorsed It? Measuring Authority Bias Across Expertise Levels in Language Modelscited by 8

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026When LLMs Sit Above Diagnostic Tools: Unrealized Complementarity in Industrial Fault Diagnosiscited by 1
  2. 2026Anatomy of LLM Sycophancy: What a Flip Rate Hidescited by 0