Protecting Copyrighted Material with Unique Identifiers in Large Language Model Training

Shuai Zhao, Linchao Zhu, Ruijie Quan +1

A primary concern regarding training large language models (LLMs) is whether they abuse copyrighted online text. With the increasing training data scale and the prevalence of LLMs…