What is it about?
Artificial intelligence learns from large sets of text and images. When those materials are copyrighted, a hard question follows: is this fair use or a violation? In this paper, I study recent disputes in China (the “Ultraman” cases) and the United States (Bartz v. Anthropic and Kadrey v. Meta) and offer a clear map. I break AI into four stages: collecting inputs, training models, generating outputs, and how people use those outputs. The law can treat each stage differently. I show where current cases line up, where they split, and what that means for artists, developers, and the market. I propose a simple rule of thumb: allow technical uses that do not replace the original or its market, and set limits on outputs that compete with the original. My goal is to support innovation while protecting human creators. I end with concrete steps for courts, platforms, and policymakers.
Featured Image
Photo by Duncan Sanchez on Unsplash
Why is it important?
I bring China and the United States into one study at the moment both courts are setting the rules for AI and copyright. I replace the vague question of “AI fair use” with four concrete questions about inputs, training, outputs, and use. This lets judges, artists, and platforms see where technical uses can be allowed and where limits are needed. It matters now because early rulings will shape markets for years. By mapping common ground across the two systems, I offer a simple path to clearer, more predictable rules that support useful AI without undercutting human creators.
Perspectives
I wrote this because I kept seeing two groups talk past each other. Creators worry about being replaced. Engineers worry about being blocked. Working between China and the United States, I wanted to offer a simple map both sides could use. The four-stage frame came out of months of reading cases and testing examples with colleagues and students. My hope is that it helps judges and platforms make clearer, faster decisions, and that it gives artists a sharper sense of what is protected and when to push back. If it also makes the AI copyright debate a little calmer and more practical, that will feel like success.
Chanhou Lou
Read the Original
This page is a summary of: Develop-Fair Use for Artificial Intelligence: A Sino-U.S. Copyright Law Comparison Based on the Ultraman, Bartz v. Anthropic, and Kadrey v. Meta Cases, January 2025, Elsevier,
DOI: 10.2139/ssrn.5458615.
You can read the full text:
Contributors
The following have contributed to this page
