Base Models Look Human To AI Detectors
As AI-generated text enters the real-world at scale, institutions increasingly use commercial AI-text detectors, especially in education and academic-integrity workflows. We report a surprising empirical finding about such systems: when evaluated by GPTZero and Pangram, generated text from base models is often judged overwhelmingly human, whereas text generated by their instruction-tuned counterparts is not. Building on this observation, we propose Humanization by Iterative Paraphrasing (HIP), a detector-agnostic pipeline that minimally fine-tunes a base model into a paraphraser and applies it iteratively. Compared with the baselines we test, HIP yields a stronger trade-off between semant...
| Authors | Yixuan Even Xu, Ziqian Zhong, Aditi Raghunathan et al. |
|---|---|
| Primary category | cs.CL |
| Published | 19 May 2026 |
| Last updated at source | 19 May 2026 |
Connected in the knowledge graph
- YixuanEvenXu/Qwen3-8B-Base-HIP-adapterModel · cited by
- YixuanEvenXu/HIP-training-and-evaluation-dataDataset · cited by
- YixuanEvenXu/Qwen3-0.6B-Base-HIP-adapterModel · cited by
- YixuanEvenXu/Qwen3-4B-Base-HIP-adapterModel · cited by
- YixuanEvenXu/Llama-3-8B-HIP-adapterModel · cited by
- YixuanEvenXu/Qwen3-1.7B-Base-HIP-adapterModel · cited by
Models and datasets referencing this paper
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.