LLM Sycophancy

LLM: Large Language Model

LLM sycophancy describes when a model agrees with or validates a user without adequate evidence, including cases where it changes a correct answer under pressure. Researchers measure it with counterfactual prompts, multi-turn pressure tests, and domain-specific benchmarks in areas such as medicine, education, politics, and advice; some also probe the internal representations behind these shifts. Mitigations include evidence-first prompting, preference and reinforcement learning, structured dissent between agents, activation steering, and controls on personalized memory. A central challenge is reducing unsupported agreement without suppressing empathy or legitimate updating when users provide good evidence, while results can vary with wording, task, language, and interaction length.

Latest papers

163 papers27 papers in the last four weeks, up 42% on the four weeks before. 0.2% of all new papers.

TrussTutor-Bench: Benchmarking large language models on solution accuracy, sycophancy, and error detection in statics
Oct 28, 2026Evangelos Apostolakis, Erfan Hamdi, Sara Jahedazad +1LLM Evaluation Language Model Reasoning Evaluation
Department of Mechanical Engineering Boston University Boston, MA 02215 · Department of Mechanical Engineering Division of Systems Engineering Boston University Boston, MA 02215
Locating Sycophancy Across Training Stages in Open Language Models
Oct 27, 2026Jiarui Yao, Shweta Parihar, Nghia Tuan Bui +2LLM Sycophancy LLM Training
Boston Children’s Hospital and Harvard Medical School · New Jersey Institute of Technology
Correction or Manipulation? Measuring Epistemic Vigilance in LLM Agents Under User Pressure
Oct 27, 2026Nardine Basta, Venus Haghighi, Dali Kaafar +1LLM Agent Evaluation LLM Agent Safety
Macquarie University, Sydney, Australia
Thinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM Agents
Oct 26, 2026Nghi D. Q. BuiLLM Sycophancy Behavioral Consistency of LLM Agents
Google
The Patient Is Always Right? Evaluating Sycophancy in Medical Diagnosis with Graded Patient Beliefs
Oct 25, 2026Viswonathan Manoranjan, Andrea Mock, Myra Cheng +4LLM Sycophancy Clinical Language Model Evaluation
University of North Carolina at Chapel Hill · Vals AI · Stanford University +2
How Much Do You Trust A Dog? Quantifying Hint-Following in Reasoning Language Models
Oct 23, 2026Anxin Guo, Chirag Pabbaraju, Sara Ahmadian +2LLM Sycophancy LLM Reliability
Northwestern University, Evanston, IL, USA · Northwestern University, Evanston, IL, USA; NSF-Simons AI Institute for the Sky (SkAI), Chicago, IL, USA · Stanford University, Stanford, CA, USA +2
Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration
Oct 22, 2026Deepthika S, Pawni Yadav, B. SankarTrust in AI LLM Reliability
Department of Electronics and Communication Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore, 641112, India · Department of Information Technology, Netaji Subhas University of Technology New Delhi, 110078, India · Department of Mechanical Engineering, Indian Institute of Science (IISc), Bangalore, 560012, India
SIM-UT: A User Simulation based Evaluation Framework for Medical Triage Chatbots
Oct 21, 2026Deepali Mittal, Dheeraj Mundhra, Yulia Sidorova +3Clinical Language Model Evaluation User Simulation
UserTrace · Cliniva · hearX Foundation +2
Consistency Is Not Validity: Self-Rewarded Moral Argumentation Makes Language Models Decisive but Pliable
Oct 21, 2026Marius De Kuthy Meurers, Aaron Parisi, Bernd BohnetMoral Reasoning in Language Models Reasoning Consistency in Language Models
Google DeepMind
Is Sycophancy a Form of In-Context Updating?
Oct 20, 2026Itai Shapira, Myra Cheng, Hadas OrgadLLM Sycophancy Belief Updating in LLMs
Harvard University · Stanford University · Kempner Institute at Harvard University
When Health Assistants Agree Too Readily: A Benchmark of Time-Critical Self-Diagnosis
Oct 20, 2026Marilyn Zhang, Tianfeng Chen, Patricia M Pechter +1Clinical Language Model Evaluation LLM Safety Benchmarks
Pareto Inc · NYU Langone
The Development of Sycophancy in Language Models: Tracing Vulnerabilities from Pretraining to Post-Training
Oct 19, 2026Haonan Huang, Chengkun CaiLLM Sycophancy Language Model Post-Training
Princeton University · University College London
No Matter Who You Are: Addressivity and Sycophancy in Large Language Models
Oct 19, 2026Seungjong Sun, Hyunmin Jun, Minsik Yoon +4LLM Sycophancy Language Model Post-Training
Department of Human-Artificial Intelligence Interaction, Sungkyunkwan University · Department of Applied Artificial Intelligence, Sungkyunkwan University · Department of Interaction Science, Sungkyunkwan University
Desirability Confirmation: Relational Sycophancy in AI Intimate Roleplay
Oct 16, 2026Huiqian Lai, Runze DingLLM Sycophancy AI Companions
Syracuse University · Swansea University
Trapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models
Oct 16, 2026Runtai Yang, Dandan Zhu, Kaiwei Zhang +1LLM Sycophancy Framing Effects in Language Models
East China Normal University · Shanghai AI Laboratory · Shanghai Jiao Tong University
Many Biases, One Circuit: Post-Training Tunes a Pretrained Cue-Routing Mechanism
Oct 16, 2026Prakhar Gupta, Terry Jingchen Zhang, Zhijing JinCognitive Bias in Language Models Circuits in Language Models
University of Michigan · Jinesis Lab, University of Toronto & Vector Institute · University of Oxford +2
Claim Framing as a Driver of LLM Sycophancy
Oct 13, 2026Jean Seo, Pepa Atanasova, Isabelle AugensteinLLM Sycophancy Framing Effects in Language Models
University of Copenhagen
Pushback in a Haystack: Language Models Respond to Factual Errors Ambiguously and Ineffectively
Oct 13, 2026Li Lucy, Myra Cheng, Svenja Guhr +7LLM Sycophancy LLM Reliability
University of Washington · University of Wisconsin-Madison · Stanford University +3
Big-Big-SycoBench: A Benchmark to Detect Sycophancy Through Model Responses to Valid vs. Invalid Arguments
Oct 11, 2026Marthe Ballon, Arne Vanhoyweghen, Seorin Kim +11LLM Sycophancy LLM Evaluation
Data Analytics Lab, Vrije Universiteit Brussel, Pleinlaan 5, 1050 Brussel, Belgium · imec-SMIT, Vrije Universiteit Brussel, Pleinlaan 9, 1050 Brussels, Belgium · School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138, USA
When Pleasing the User Compromises Care: AI Sycophancy in Caregiver-Mediated Healthcare
Oct 9, 2026Priyadarsi Mishra, Arjun Beniwal, Ziyu Yi +2LLM Sycophancy Human-AI Decision Making
Stanford University · Texas A&M University
The AI's Poker Face: Contrasting Persuasion and Sycophancy in Voice-Based Multi-Turn Conversations
Oct 9, 2026Ilia Frantzi, Bernie Hogan, Max M. Lang +1Persuasion in Language Models LLM Sycophancy
University of Munich, Germany · University of Oxford, UK · GESIS - Leibniz-Institute for the Social Sciences, Germany
BeliefScope: Diagnosing Evidence-Driven Revision and Pressure-Induced Shifts in Large Language Models
Oct 8, 2026Shuai Guo, Yidong CuiBelief Updating in LLMs LLM Auditing
Beijing University of Posts and Telecommunications Beijing, China
Defense-in-Depth for LLMs: Evaluating Memory Gates Against Activation-Induced and Memory-Induced Sycophancy
Oct 5, 2026Ritvij Sharma, Russell Dlugosz, Ryan Zhou +1LLM Sycophancy Memory-Augmented Language Models
Urbana CS Club
Anatomy of LLM Sycophancy: What a Flip Rate Hides
Oct 5, 2026Haonan HuangLLM Evaluation LLM Sycophancy
Princeton University
Memadapter: Counterfactual Adaptation Against Memory-induced Sycophancy
Oct 4, 2026Ruqing Ning, Haibo Meng, Zhishang Xiang +4LLM Sycophancy LLM Agent Memory
Jilin University · Xiamen University
Suppressing Pressure, Amplifying Evidence: Self-Guided Attention Steering to Mitigate Sycophancy and Stubbornness
Oct 3, 2026Yinghao He, Mengyu Xu, Haixiang Sun +8LLM Sycophancy Language Model Robustness
Purdue University · The Ohio State University · Analogy AI, Inc. +2
Mitigating Social Sycophancy via Pluralistic Preference Optimization
Oct 1, 2026Stephane Hatgis-Kessell, Myra Cheng, Xiaoxuan Hou +4LLM Sycophancy Preference Optimization
Department of Computer Science, Stanford University · Amazon · Department of Computer Science, University of Washington
FIGS: Evaluating Multi-Turn Sycophancy Without Penalizing Empathy
Sep 30, 2026Sidharth Pulipaka, Ruta Binkyte, Ivaxi Sheth +1LLM Sycophancy Multi-Turn Dialogue Evaluation
Independent · German Research Center for Artificial Intelligence (DFKI) · ELLIS Institute Tübingen +2
Beyond the Sycophancy Score: How Task, Model, and Pressure Shape LLM Yielding
Sep 30, 2026Guang Yang, Homa Hosseinmardi, Fengchen Liu +1LLM Sycophancy LLM Reliability
OASIS Lab, University of California, Los Angeles · University of California, Berkeley
Authority Bias in Language Models: Source Deference and User Agreement Are Not Interchangeable
Sep 29, 2026Abhinav Rajeev Kumar, Paras ChopraLLM Sycophancy LLM Reliability
Lossfunk

Oct 28, 2026J/K move · Enter open · S save

TrussTutor-Bench: Benchmarking large language models on solution accuracy, sycophancy, and error detection in statics

Evangelos Apostolakis, Erfan Hamdi, Sara Jahedazad, Emma Lejeune

Department of Mechanical Engineering Boston University Boston, MA 02215 · Department of Mechanical Engineering Division of Systems Engineering Boston University Boston, MA 02215

LLM Evaluation Language Model Reasoning Evaluation LLM Sycophancy Language Model Error Detection

Open paper AskPDF arXiv

Large language models (LLMs) have rapidly become a default source of help for students in science and engineering courses. In foundational engineering mechanics courses such as statics, students routinely use these tools to check answers, work through partial solutions, and ask for explanations. Effective tutoring in this setting requires more than a correct final answer. A model must also confirm correct work when a student doubts it, and identify errors in flawed work. Existing benchmarks evaluate LLMs primarily on final-answer accuracy, so it is unclear whether models that can solve mechanics problems also have the other technical capabilities that tutoring depends on. Here we introduce TrussTutor-Bench, a benchmark of 200 verified two-dimensional truss problems paired with automatically generated student solutions, and use it to evaluate thirteen LLMs on solution accuracy, sycophancy, and error detection. Baseline accuracy ranged from 64.5% to 97.0%. Every model detected nearly all injected errors in synthetic flawed student work (90.4% to 100%), so error detection did not distinguish models under these clean conditions. Sycophancy did. When a student insisted that correct work was wrong, models rejected correct solutions to problems they had themselves solved in 1.6% to 50.0% of cases. These results show that performance varies widely across models: the leading models solve, check, and defend student work nearly perfectly, while others are far less reliable, particularly when a student doubts correct work. TrussTutor-Bench provides a verified, extensible standard for determining which models are technically reliable enough to put in front of students. Follow-up work must now focus on teaching rather than technical capability, drawing on learning science to determine whether LLM help leads students to understand the material or simply to reach the right answer.