Redirecting to original paper in 30 seconds...

Click below to go immediately or wait for automatic redirect

arxiv_cl 95% Match Research Paper AI Researchers,Content Moderators,Educators,Cybersecurity Professionals 4 weeks ago

FAID: Fine-Grained AI-Generated Text Detection Using Multi-Task Auxiliary and Multi-Level Contrastive Learning

large-language-models › evaluation
📄 Abstract

Abstract: The growing collaboration between humans and AI models in generative tasks has introduced new challenges in distinguishing between human-written, LLM-generated, and human--LLM collaborative texts. In this work, we collect a multilingual, multi-domain, multi-generator dataset FAIDSet. We further introduce a fine-grained detection framework FAID to classify text into these three categories, and also to identify the underlying LLM family of the generator. Unlike existing binary classifiers, FAID is built to capture both authorship and model-specific characteristics. Our method combines multi-level contrastive learning with multi-task auxiliary classification to learn subtle stylistic cues. By modeling LLM families as distinct stylistic entities, we incorporate an adaptation to address distributional shifts without retraining for unseen data. Our experimental results demonstrate that FAID outperforms several baselines, particularly enhancing the generalization accuracy on unseen domains and new LLMs, thus offering a potential solution for improving transparency and accountability in AI-assisted writing.

Key Contributions

FAID is a fine-grained AI-generated text detection framework that classifies text into human-written, LLM-generated, or collaborative categories, and identifies the LLM family. It uses multi-level contrastive learning and multi-task auxiliary classification to capture subtle stylistic cues and incorporates adaptation for distributional shifts.

Business Value

Essential for maintaining academic integrity, combating misinformation, and ensuring authenticity in digital content. It supports platforms in moderating content and verifying authorship.