Agents Don't Just Agree, They Remember: Benchmarking Persistent Sycophancy in Self-Improving Personal Agents
cs.AI
Submitted: 2026-07-12
Updated: 2026-09-27
Code: https://github.com/henrymao2004/agent-sycophancy
Project page: https://henrymao2004.github.io/agent-sycophancy
Terminology
Sources
- Beyond Social Pressure: Benchmarking Epistemic Attack in Large Language Models
- SWAY: A Counterfactual Computational Linguistic Approach to Measuring and Mitigating Sycophancy
- Memory for Autonomous LLM Agents:Mechanisms, Evaluation, and Emerging Frontiers
- Sycophantic Anchors: Localizing and Quantifying User Agreement in Reasoning Models
- Good Arguments Against the People Pleasers: How Reasoning Mitigates (Yet Masks) LLM Sycophancy
- User Preference Modeling for Conversational LLM Agents: Weak Rewards from Retrieval-Augmented Interaction
- MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks
- OP-Bench: Benchmarking Over-Personalization for Memory-Augmented Personalized Conversational Agents
- Personalization Increases Affective Alignment but Has Role-Dependent Effects on Epistemic Independence in LLMs
- Learning Personalized Agents from Human Feedback
- A Survey on Long-Term Memory Security in LLM Agents: Attacks, Defenses, and Governance Across the Memory Lifecycle
- Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems
- Measuring Opinion Bias and Sycophancy via LLM-based Persuasion
- BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs
- PersistBench: When Should Long-Term Memories Be Forgotten by LLMs?
- PENDULUM: A Benchmark for Assessing Sycophancy in Multimodal Large Language Models
- The Silicon Mirror: Dynamic Behavioral Gating for Anti-Sycophancy in LLM Agents
- Mem2ActBench: A Benchmark for Evaluating Long-Term Memory Utilization in Task-Oriented Autonomous Agents
- Evaluating Memory Structure in LLM Agents
- Political Bias Audits of LLMs Capture Sycophancy to the Inferred Auditor
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection