Protect or Promote?: Strategic Framing and Stakeholder Tensions in the AI Copyright Debate

• In Event: Regulating and Reframing AI Policy

Abstract

Can copyright encourage innovation, or does it get in the way—especially in the age of AI? This question has been debated for a long time, and it has become even more important as generative AI challenges traditional ideas about authorship and ownership. Despite this urgency, there are still no clear laws or policies that explain how copyright should apply to these new technologies. This study explores how different stakeholders—such as tech companies, creators, users, and policymakers—take positions on key copyright issues. We focus on three main areas: whether copyrighted material can be used to train AI (input), whether AI-generated content should be protected by copyright (output), and whether the AI model itself can be copyrighted (model). To analyze these perspectives, we used a dataset of 10,373 public comments submitted to the U.S. Copyright Office as part of the 2023–2024 study on Artificial Intelligence and Copyright. We applied BERT topic modeling and TF-IDF methods to uncover patterns in stakeholder positions. Our results show a clear pattern. Tech companies usually support fair use of training data but also want strong copyright protection for the content produced by their AI. On the other hand, creators and users consistently oppose the unlicensed use of copyrighted materials and giving copyright to AI-generated content. These differences are not just casual disagreements. They reveal deeper strategic conflicts about how copyright is used—and who benefits from it.

Author

• Soojung Crystal Chun, Georgia Institute of Technology
  Presenting Author