arXiv：cs.CL· Jicheng Zhou, Kahim Wong, Jialong Wang, Jiantao Zhou·· 3 小时前AI 评分32
EASE：用熵自适应分布塑形规避 AI 生成文本检测器
EASE: Entropy-Adaptive Distribution Shaping for Evading AI-generated Text Detectors
AI 导读
研究者提出 EASE（Entropy-Adaptive Distribution Shaping for Evasion），一个免训练、与检测器无关的框架，通过源 LLM 的预测熵自适应调整 logit 扰动和采样温度来规避 AI 生成文本检测器。在三个源 LLM 和多个检测器上的实验显示检测性能一致下降，且文本质量与推理开销的损失可忽略。该方法无需检测器反馈，也无需模型微调。
正文
Abstract:AI-generated text (AIGT) detection can be sensitive to the decoding choices of the source large language model (LLM). We observe that perturbing next-token logits or adjusting sampling temperature can reduce detection performance, providing a clear signal of detector vulnerability to decoding-time distribution changes. Building on this observation, we propose EASE (Entropy-Adaptive Distribution Shaping for Evasion), a training-free and detector-agnostic framework for evading AIGT detectors. EASE computes predictive entropy directly from the source LLM's next-token distribution and uses it to adapt both logit perturbation and sampling temperature, without detector feedback or model fine-tuning. Experiments across three source LLMs and multiple detectors demonstrate consistent reductions in detection performance, with negligible degradation in text quality and negligible inference overhead.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.09976 [cs.CL] |
| (or arXiv:2610.09976v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09976 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jicheng Zhou [view email]
[v1]
Wed, 7 Oct 2026 12:41:19 UTC (229 KB)
来源：arXiv：cs.CL · arxiv.org