现有大模型文本检测器在真实场景中不可靠，新方法提升鲁棒性但仍有局限
Can We Trust LLM Detectors?
- 用对比学习提取文本风格特征，增强检测区分能力
- 监督式检测器在域外数据上性能骤降超过40%
- 无需训练的检测方法对代理数据高度敏感，适合特定场景
大语言模型的快速应用增加了对可靠AI文本检测的需求，但现有检测器在受控基准之外常失效。我们系统评估了两种主流范式（无训练与有监督），发现两者在分布偏移、未知生成器及简单风格扰动下均表现脆弱。为此，我们提出一种监督对比学习（SCL）框架，学习具有判别性的风格嵌入。实验表明，尽管有监督检测器在域内表现优异，但在域外性能急剧下降超过40%；而无训练方法对代理选择仍高度敏感。总体而言，我们的结果揭示了构建领域无关检测器的根本挑战。代码已开源：https://github.com/HARSHITJAIS14/DetectAI
原文摘要 · Abstract (English)
The rapid adoption of LLMs has increased the need for reliable AI text detection, yet existing detectors often fail outside controlled benchmarks. We systematically evaluate 2 dominant paradigms (training-free and supervised) and show that both are brittle under distribution shift, unseen generators, and simple stylistic perturbations. To address these limitations, we propose a supervised contrastive learning (SCL) framework that learns discriminative style embeddings. Experiments show that while supervised detectors excel in-domain, they degrade sharply out-of-domain, and training-free methods remain highly sensitive to proxy choice. Overall, our results expose fundamental challenges in building domain-agnostic detectors. Our code is available at: https://github.com/HARSHITJAIS14/DetectAI
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。