THE HEART OF THE MATTER: COPYRIGHT, AI TRAINING AND LLM

71 J. COPYRIGHT SOC’Y, 482, (2024)

Daniel Gervais Vanderbilt University

Haralambos Marmanis Copyright Clearance Center, EVP & CTO

Noam Shemtov Centre for Commercial Law, Queen Mary University of London

Catherine Zaller Rowland Copyright Clearance Center, General Counsel

Abstract

This article explores the intricate relationship between copyright law and artificial intelligence, including large language models (LLMs). It begins with a detailed technical overview of LLM functionality, including tokenization, word embeddings, and the various stages of LLM development. The authors then delve into the copyright implications of using protected works for both training LLMs and generating outputs.