DetectRL-X: Towards Reliable Multilingual and Real-World LLM-Generated Text Detection

Junchao Wu, Yefeng Liu, Chenyu Zhu, Hao Zhang, Zeyu Wu, Tianqi Shi, Yichao Du, Longyue Wang, Weihua Luo, Jinsong Su, Derek F. Wong

arXiv:2605.15518 · 2026-05-18 공개 · arXiv · PDF

text-detection llm-generated-text paraphrasing-attacks real-world-applications language-specific-detectors detector-evaluation llm-misuse text-perturbation

Abstract

The effective detection and governance of Large Language Model (LLM) generated content has become increasingly critical due to the growing risk of misuse. Despite the impressive performance of existing detectors, their reliability and potential in multilingual, real-world scenarios remain largely underexplored. In this study, we introduce DetectRL-X, a comprehensive multilingual benchmark designed to evaluate advanced detectors across 8 dimensions. The benchmark encompasses 8 languages commonly used in commercial contexts and collects human-written texts from 6 domains highly susceptible to LLM misuse. To better aligned with real-world applications, We create LLM-generated texts using 4 popular commercial LLMs, and include typical AI-assisted writing operations such as polishing, expanding, and condensing to capture authentic usage patterns. Furthermore, we develop a multilingual framework for paraphrasing and perturbation attacks to simulate diverse human modifications and writing noise, enabling stress testing of detectors across languages. Experimental results on DetectRL-X reveal the strengths and limitations of current state-of-the-art detectors when applied to diverse linguistic resources. We further analyze how domains, generators, attack strategies, text length, and refinement operations influence performance in different languages, underscoring DetectRL-X as an effective benchmark for strengthening multilingual and language-specific detectors.

한국어 요약

📋 한 줄 요약

**[NLP 평가 / LLM 텍스트 탐지]** 8개 언어·6개 도메인·4개 상용 LLM·다양한 공격 시나리오를 포괄하는 다국어 실사용 LLM 생성 텍스트 탐지 벤치마크 DetectRL-X 공개.

🎯 핵심 기여도

💡 핵심 아이디어

실제 LLM 오남용은 단일 언어·단일 도메인이 아니라 다국어·다도메인 + 사람의 후처리를 포함하는 복합 시나리오에서 발생하므로, 탐지 평가도 이 복합성을 동일하게 재현해야 한다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: LLM 텍스트 탐지 연구의 평가 표준을 단일 언어·합성 시나리오에서 실세계 다국어·다공격 시나리오로 확장. **한계**: 8개 언어와 6개 도메인 외 자원이 부족한 언어·전문 도메인으로의 일반화는 후속 과제.

🚀 실용적 활용