arXiv cs.AI· Suraj Ranganath, Atharv Ramesh·· 12 小时前AI 评分47
StealthRL：用强化学习改写攻击让 AI 文本检测器集体失效
StealthRL: Reinforcement Learning Paraphrase Attacks for Multi-Detector Evasion of AI-Text Detectors
AI 导读
StealthRL 是一个强化学习框架，通过训练面向检测器集成系统的改写策略并约束语义保真度，对 AI 文本检测器发起自适应改写攻击。在 RoBERTa、Fast-DetectGPT、Binoculars 和 MAGE 上，该攻击将平均 TPR@1%FPR 压至 0.024，同时保留可读文本与语义重叠，并能迁移到训练中未见的检测器。代码与训练好的模型 checkpoint 已公开。
正文 · 原文
Abstract:AI-text detectors are increasingly used in high-stakes settings, but their robustness to meaning-preserving adversarial rewriting remains unclear. We introduce StealthRL, a reinforcement learning framework for stress-testing detectors with adaptive paraphrase attacks. StealthRL trains a paraphrase policy against a detector ensemble while enforcing semantic fidelity, then evaluates transfer to held-out detector families on the full filtered MAGE test pool. Across RoBERTa, Fast-DetectGPT, Binoculars, and MAGE, the resulting attack sharply degrades detection performance, driving mean TPR@1%FPR down to 0.024 while retaining nontrivial semantic overlap and readable prose. The attack also transfers to detectors not used during training, indicating a broader robustness failure rather than overfitting to a single target model. We complement the main results with detector-score analysis, repeated-generation stability checks, and automatic and judge-based quality evaluation. Our findings suggest that current AI-text detectors remain brittle under adaptive paraphrasing pressure and motivate robustness evaluation at deployment-relevant operating points. Our code and the trained model checkpoint are publicly available.
| Comments: | Long paper at the First Workshop on Pluralistic Value Alignment of LLMs (PlurVA-LLM 2026) @ AACL-IJCNLP 2026. 16 pages. Code: this https URL |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) |
| Cite as: | arXiv:2602.08934 [cs.LG] |
| (or arXiv:2602.08934v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2602.08934 arXiv-issued DOI via DataCite |
Submission history
From: Suraj Ranganath [view email]
[v1]
Mon, 9 Feb 2026 17:33:46 UTC (13,020 KB)
[v2]
Thu, 19 Mar 2026 23:19:31 UTC (8,988 KB)
[v3]
Fri, 23 Oct 2026 08:21:02 UTC (755 KB)
来源：arXiv cs.AI · arxiv.org