Fight Poison with Poison: Enhancing Robustness in Few-shot Machine-Generated Text Detection with Adversarial Training

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Duan, Wenjing, Zhou, Qi, Li, Yuanfan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913085764141056
author Duan, Wenjing
Zhou, Qi
Li, Yuanfan
author_facet Duan, Wenjing
Zhou, Qi
Li, Yuanfan
contents Machine-generated text (MGT) detection is critical for regulating online information ecosystems, yet existing detectors often underperform in few-shot settings and remain vulnerable to adversarial, humanizing attacks. To build accurate and robust detectors under limited supervision, we adopt a threat-modeling perspective and study detector vulnerabilities from an attacker's viewpoint under an output-only black-box setting. Motivated by this perspective, we propose RAG-GuidEd Attacker Strengthens ConTrastive Few-shot Detector (REACT), an adversarial training framework that improves both few-shot detection performance and robustness against attacks. REACT couples a humanization-oriented attacker with a target detector: the attacker leverages retrieval-augmented generation (RAG) to craft highly human-like adversarial examples to evade detection, while the detector learns from these adversaries with a contrastive objective to stabilize few-shot representation learning and enhance robustness. We alternately update the attacker and the detector to enable their co-evolution. Experiments on 4 datasets with 4 shot sizes and 3 random seeds show that REACT improves average detection F1 by 4.95 points over 8 state-of-the-art (SOTA) detectors and reduces the average attack success rate (ASR) under 4 strong attacks by 3.66 percentage points.
format Preprint
id arxiv_https___arxiv_org_abs_2605_02374
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Fight Poison with Poison: Enhancing Robustness in Few-shot Machine-Generated Text Detection with Adversarial Training
Duan, Wenjing
Zhou, Qi
Li, Yuanfan
Cryptography and Security
Computation and Language
Machine-generated text (MGT) detection is critical for regulating online information ecosystems, yet existing detectors often underperform in few-shot settings and remain vulnerable to adversarial, humanizing attacks. To build accurate and robust detectors under limited supervision, we adopt a threat-modeling perspective and study detector vulnerabilities from an attacker's viewpoint under an output-only black-box setting. Motivated by this perspective, we propose RAG-GuidEd Attacker Strengthens ConTrastive Few-shot Detector (REACT), an adversarial training framework that improves both few-shot detection performance and robustness against attacks. REACT couples a humanization-oriented attacker with a target detector: the attacker leverages retrieval-augmented generation (RAG) to craft highly human-like adversarial examples to evade detection, while the detector learns from these adversaries with a contrastive objective to stabilize few-shot representation learning and enhance robustness. We alternately update the attacker and the detector to enable their co-evolution. Experiments on 4 datasets with 4 shot sizes and 3 random seeds show that REACT improves average detection F1 by 4.95 points over 8 state-of-the-art (SOTA) detectors and reduces the average attack success rate (ASR) under 4 strong attacks by 3.66 percentage points.
title Fight Poison with Poison: Enhancing Robustness in Few-shot Machine-Generated Text Detection with Adversarial Training
topic Cryptography and Security
Computation and Language
url https://arxiv.org/abs/2605.02374