English

PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning

Computation and Language 2025-06-19 v1 Computers and Society

Abstract

With the popularity of large language models (LLMs), undesirable societal problems like misinformation production and academic misconduct have been more severe, making LLM-generated text detection now of unprecedented importance. Although existing methods have made remarkable progress, a new challenge posed by text from privately tuned LLMs remains underexplored. Users could easily possess private LLMs by fine-tuning an open-source one with private corpora, resulting in a significant performance drop of existing detectors in practice. To address this issue, we propose PhantomHunter, an LLM-generated text detector specialized for detecting text from unseen, privately-tuned LLMs. Its family-aware learning framework captures family-level traits shared across the base models and their derivatives, instead of memorizing individual characteristics. Experiments on data from LLaMA, Gemma, and Mistral families show its superiority over 7 baselines and 3 industrial services, with F1 scores of over 96%.

Keywords

Cite

@article{arxiv.2506.15683,
  title  = {PhantomHunter: Detecting Unseen Privately-Tuned LLM-Generated Text via Family-Aware Learning},
  author = {Yuhui Shi and Yehan Yang and Qiang Sheng and Hao Mi and Beizhe Hu and Chaoxi Xu and Juan Cao},
  journal= {arXiv preprint arXiv:2506.15683},
  year   = {2025}
}

Comments

17 pages, 3 figures, 6 tables