Author

Natesh Reddy

Publication Date

Spring 2025

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

Mark Stamp

Second Advisor

Thomas Austin

Third Advisor

Genya Ishigaki

Keywords

Sequence-to-Sequence Models, Style Transfer, Natural Language Processing (NLP), Large Language Models (LLMs), Adversarial Text Transformation

Abstract

With the advancements in Large Language Models (LLMs) such as ChatGPT, the boundary between text written by a human and that created by AI has become blurred. This poses a threat to systems designed to detect AI-generated content. In this work, we adopt a novel strategy to adversarially transform GPT-generated text using sequence-to-sequence (Seq2Seq) models, to make the text more human-like. We concentrate on improving GPT-generated sentences by including significant linguistic, structural, and semantic components that are typical of human-authored text. The goal is to then use this transformed data to train a more robust detection model. Our experiments show that our Seq2Seq approach makes it difficult for classifiers to detect AI-generated text. On the other hand, once retrained on this augmented data, models become more robust with improved accuracy in distinguishing AI-generated from human-generated content. This work adds to the ongoing conversation of text transformation as a tool for both attack (in the sense of defeating classification models) and defense (in the sense of improved classifiers), thereby advancing our understanding of AI-generated text detection.

Available for download on Monday, May 25, 2026

Share

COinS