研究：LLM 在用户引导下更易输出历史修正主义内容，HistoricalMisinfo 基准发布

Do LLMs Rewrite History? Detecting Historical Revisionism in Large Language Models

AI 导读

研究者构建 HistoricalMisinfo 数据集，覆盖 45 个国家和地区的 500 起争议历史事件，每起事件配事实版与有据可查的修正主义版两种参考叙事，并在 11 种提示场景下实例化。采用 LLM-as-a-judge 协议评测多种模型架构：中性提示下模型总体贴近事实参考，但一旦提示明确要求修正主义内容，所有模型的修正主义得分均大幅上升，显示对用户引导的修正主义框架抵抗力有限。

正文 · 原文

View PDF HTML (experimental)

Abstract:Large language models (LLMs) are increasingly used to disseminate information, yet their handling of contested historical events and politically charged narratives remains understudied. We introduce HistoricalMisinfo, a dataset of 500 contested events from 45 countries and regions, each paired with a factual and a documented revisionist reference narrative, instantiated across 11 prompt scenarios reflecting common real-world usage. Using an LLM-as-a-judge protocol, we evaluate multiple model architectures under neutral prompts and robustness prompts that explicitly request revisionist content. Under neutral prompts, models generally align with factual references, though alignment varies by scenario and model. Robustness prompting reveals a consistent and sharp increase in revisionism scores across all models, indicating limited resistance to user-driven revisionist framing. HistoricalMisinfo offers a practical benchmark for evaluating robustness to historical revisionism in AI systems. Code and dataset: this https URL

Comments:	EMNLP 2026
Subjects:	Computers and Society (cs.CY)
Cite as:	arXiv:2602.17433 [cs.CY]
	(or arXiv:2602.17433v3 [cs.CY] for this version)
	https://doi.org/10.48550/arXiv.2602.17433 arXiv-issued DOI via DataCite

Submission history

From: Joeun Yook [view email]
[v1] Thu, 19 Feb 2026 15:05:10 UTC (3,508 KB)
[v2] Sun, 22 Feb 2026 18:42:06 UTC (3,508 KB)
[v3] Mon, 26 Oct 2026 23:37:16 UTC (3,778 KB)

来源：arXiv cs.CY · arxiv.org