Authorship Obfuscation in Multilingual Machine-Generated Text Detection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Macko, Dominik, Moro, Robert, Uchendu, Adaku, Srba, Ivan, Lucas, Jason Samuel, Yamashita, Michiharu, Tripto, Nafis Irtiza, Lee, Dongwon, Simko, Jakub, Bielikova, Maria
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909410979217408
author Macko, Dominik
Moro, Robert
Uchendu, Adaku
Srba, Ivan
Lucas, Jason Samuel
Yamashita, Michiharu
Tripto, Nafis Irtiza
Lee, Dongwon
Simko, Jakub
Bielikova, Maria
author_facet Macko, Dominik
Moro, Robert
Uchendu, Adaku
Srba, Ivan
Lucas, Jason Samuel
Yamashita, Michiharu
Tripto, Nafis Irtiza
Lee, Dongwon
Simko, Jakub
Bielikova, Maria
contents High-quality text generation capability of recent Large Language Models (LLMs) causes concerns about their misuse (e.g., in massive generation/spread of disinformation). Machine-generated text (MGT) detection is important to cope with such threats. However, it is susceptible to authorship obfuscation (AO) methods, such as paraphrasing, which can cause MGTs to evade detection. So far, this was evaluated only in monolingual settings. Thus, the susceptibility of recently proposed multilingual detectors is still unknown. We fill this gap by comprehensively benchmarking the performance of 10 well-known AO methods, attacking 37 MGT detection methods against MGTs in 11 languages (i.e., 10 $\times$ 37 $\times$ 11 = 4,070 combinations). We also evaluate the effect of data augmentation on adversarial robustness using obfuscated texts. The results indicate that all tested AO methods can cause evasion of automated detection in all tested languages, where homoglyph attacks are especially successful. However, some of the AO methods severely damaged the text, making it no longer readable or easily recognizable by humans (e.g., changed language, weird characters).
format Preprint
id arxiv_https___arxiv_org_abs_2401_07867
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Authorship Obfuscation in Multilingual Machine-Generated Text Detection
Macko, Dominik
Moro, Robert
Uchendu, Adaku
Srba, Ivan
Lucas, Jason Samuel
Yamashita, Michiharu
Tripto, Nafis Irtiza
Lee, Dongwon
Simko, Jakub
Bielikova, Maria
Computation and Language
High-quality text generation capability of recent Large Language Models (LLMs) causes concerns about their misuse (e.g., in massive generation/spread of disinformation). Machine-generated text (MGT) detection is important to cope with such threats. However, it is susceptible to authorship obfuscation (AO) methods, such as paraphrasing, which can cause MGTs to evade detection. So far, this was evaluated only in monolingual settings. Thus, the susceptibility of recently proposed multilingual detectors is still unknown. We fill this gap by comprehensively benchmarking the performance of 10 well-known AO methods, attacking 37 MGT detection methods against MGTs in 11 languages (i.e., 10 $\times$ 37 $\times$ 11 = 4,070 combinations). We also evaluate the effect of data augmentation on adversarial robustness using obfuscated texts. The results indicate that all tested AO methods can cause evasion of automated detection in all tested languages, where homoglyph attacks are especially successful. However, some of the AO methods severely damaged the text, making it no longer readable or easily recognizable by humans (e.g., changed language, weird characters).
title Authorship Obfuscation in Multilingual Machine-Generated Text Detection
topic Computation and Language
url https://arxiv.org/abs/2401.07867