ResearchResearch paperLarge Language Models1 source · Oct 16, 2026

Trapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models

Large language models (LLMs) are known to exhibit sycophancy by adjusting their responses toward users' expressed views.

However, real users often justify their opinions through reasons, personal experiences, causal explanations, and broader narratives, raising the question of whether such narratives can further steer model judgments.
We construct a controlled benchmark of 500 debatable questions across ten domains, where the user stance is held fixed while its justification is systematically varied without introducing new task-relevant evidence.
We evaluate seven LLMs and observe substantial differences in narrative-induced agreement.
The effect is not monotonic with narrative elaboration: personal experience can reduce agreement, whereas causal framing and worldview-based synthesis increase it.

[1]Trapped in Your Story: Evaluating Narrative-Induced Sycophancy in Large Language Models
arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 02:26 PM
Large language models (LLMs) are known to exhibit sycophancy by adjusting their responses toward users' expressed views.
However, real users often justify their opinions through reasons, personal experiences, causal explanations, and broader narratives, raising the question of whether such narratives can further steer model judgments.

Extractive summary: sentences quoted from the sources.

Before this