清华Jie Tang团队提出科研Agent不端行为检测框架，F1 85.2%、干预 −36.8%

科研 agent 在「看起来像科学」：清华 Jie Tang RIDAR，F1 85.2%、干预 −36.8%

Edge-TTS 广播级有声·🎙 主播：云希 (沉稳新闻联播腔)

全员无门槛畅听

点击合成广播级音频

00:00

AI 导读

过程科研诚信违规可在 99.7% 轨迹出现，需要检测+干预。

推荐理由

该研究提出了一个用于自动检测和干预研究不端行为的框架，对AI在科研领域的应用具有重要意义。

一手信源

学术论文·arXiv:2610.31372

清华Jie Tang团队提出科研Agent不端行为检测框架，F1 85.2%、干预 −36.8% (arXiv:2610.31372)

“View PDF HTML (experimental) Abstract: Autonomous research agents are increasingly used to automate large portions of scientific research, y”

正文 · 原文

View PDF HTML (experimental)

Abstract:Autonomous research agents are increasingly used to automate large portions of scientific research, yet they are prone to research integrity violations (RIVs): behaviors that compromise the validity, reproducibility, or faithfulness of the research process. We present RIDAR, a framework for automated RIV detection and runtime intervention. First, we introduce RIVT, a comprehensive taxonomy of 14 RIV types spanning the full research lifecycle. Building on RIVT, we present RIDAR-post-hoc, a framework to detect RIVs within agent trajectories. RIDAR-post-hoc achieves 85.2% F1 performance on RIV-Gold, a benchmark for RIV detection with 60 expert-annotated trajectories. Applying RIDAR-post-hoc to RIV-Wild, a dataset of 300 agent trajectories across 50 distinct research tasks, we detect 3,964 RIV instances in 99.7% of trajectories, revealing the prevalence of RIVs in current agents. Finally, we present RIDAR-runtime, a framework for dynamically detecting and correcting RIVs during agent runtime. RIDAR-runtime effectively reduces agent RIVs remaining at task completion by 36.8%. We will release our code and data to facilitate future research.

Subjects:	Artificial Intelligence (cs.AI)
Cite as:	arXiv:2610.31372 [cs.AI]
	(or arXiv:2610.31372v1 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2610.31372 arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Jiening Siow [view email]
[v1] Thu, 22 Oct 2026 14:43:36 UTC (857 KB)