arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.11638cs.CL
收藏

绊脚石:压力测试机器生成文本检测器在攻击下的鲁棒性

Stumbling Blocks: Stress Testing the Robustness of Machine-Generated Text Detectors Under Attacks

Paul G. Allen School of Computer Science & Engineering, University of Washington(华盛顿大学保罗·G·艾伦计算机科学与工程学院);Xi’an Jiaotong University(西安交通大学);Johns Hopkins University(约翰斯·霍普金斯大学);Peking University(北京大学)

Yichen Wang, Shangbin Feng, Abe Bohan Hou, Xiao Pu, Chao Shen, Xiaoming Liu, Yulia Tsvetkov, Tianxing He

AI总结:

本研究在真实场景下压力测试了主流机器生成文本检测器对编辑、改写、提示和协同生成等攻击的鲁棒性,发现现有检测器均存在漏洞,平均性能下降35%,并提出了初步补丁以提升鲁棒性。

摘要:

大型语言模型(LLMs)的广泛使用增加了对检测机器生成文本以防止滥用方法的需求。我们研究的目的是在真实场景下压力测试检测器对恶意攻击的鲁棒性。我们全面研究了流行的机器生成文本检测器在多种攻击类别下的鲁棒性:编辑、改写、提示和协同生成。我们的攻击假设对生成器LLMs的访问受限,并比较了检测器在不同预算水平下对不同攻击的性能。我们的实验表明,几乎没有现有检测器在所有攻击下保持鲁棒,所有检测器都表现出不同漏洞。平均所有检测器,性能在所有攻击中下降35%。此外,我们调查了这些缺陷背后的原因,并提出了开箱即用的初步补丁以提高鲁棒性。

英文摘要:

The widespread use of large language models (LLMs) is increasing the demand for methods that detect machine-generated text to prevent misuse. The goal of our study is to stress test the detectors' robustness to malicious attacks under realistic scenarios. We comprehensively study the robustness of popular machine-generated text detectors under attacks from diverse categories: editing, paraphrasing, prompting, and co-generating. Our attacks assume limited access to the generator LLMs, and we compare the performance of detectors on different attacks under different budget levels. Our experiments reveal that almost none of the existing detectors remain robust under all the attacks, and all detectors exhibit different loopholes. Averaging all detectors, the performance drops by 35% across all attacks. Further, we investigate the reasons behind these defects and propose initial out-of-the-box patches to improve robustness.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑