Artificial Intelligence and the Evolution of Copyright Law: A Comparative Legal Analysis Across Global Entertainment Jurisdictions
Written by: Nathanael Jackson
January 2nd, 2026
AI, Copyright, and the Global Divide Over Creative Rights
When AI generates a song, image, or screenplay, the copyright questions begin long before the finished work appears.
A song can reflect years of practice. An illustration can carry an artist’s distinctive perspective. A screenplay can emerge from months of writing and revision. Generative artificial intelligence can produce similar forms of expression in seconds—but that speed brings complicated questions about the creative work used to develop these systems.
What happens when copyrighted material becomes training data? When does AI-assisted work qualify for copyright protection? And how should creators understand their rights when the same technology operates across countries with different legal rules?
My research examines these questions through a comparison of copyright frameworks in the United States, the United Kingdom, South Korea, and China. Rather than centering on unresolved lawsuits, the paper looks at the design of copyright law: how national systems structure authorship, exceptions, and protections for creative work.
Its central concern is not simply whether AI can produce something that appears creative. It is whether existing copyright frameworks can meaningfully protect human creative agency in an increasingly interconnected digital environment.
Three Questions Behind the AI Copyright Debate:
Public discussions often treat “AI and copyright” as a single issue. In practice, several distinct questions are involved.
The first concerns training material. Developing an AI model may involve copying and processing text, images, music, or other works. Those materials can include copyrighted works, licensed content, and public-domain material. Whether a particular use requires permission depends on the rights involved, the applicable law, and any relevant exceptions.
The second concerns authorship and protection. When someone uses AI in a creative process, which human contributions qualify for copyright? A person’s original writing, revisions, or creative arrangement may raise different questions from material generated with little meaningful human creative input.
The third concerns the resulting output’s relationship to existing works. An output’s eligibility for copyright protection is separate from whether it reproduces protected expression from someone else’s work.
These distinctions matter. Lawful training does not automatically make an output copyrightable or non-infringing. Likewise, copyright protection for a human contribution does not establish that the underlying training material was lawfully used.
One Technology, Different Copyright Rules:
AI systems can operate internationally, but copyright remains largely territorial. International treaties establish shared obligations, while national laws determine many of the details.
The four jurisdictions examined in my paper illustrate different approaches to those details:
United States: Fair use offers a flexible, case-specific framework for assessing certain unauthorized uses of copyrighted works. That flexibility can accommodate unfamiliar technologies, but it does not provide an automatic exemption for AI training. Outcomes depend on the particular use and the statutory factors.
United Kingdom: Fair dealing and specific statutory exceptions organize permitted uses around defined purposes and conditions. Its text-and-data-mining exception for non-commercial research makes the purpose of the activity and lawful access to the material particularly important. The UK also has provisions for certain computer-generated works, complicating any claim that every jurisdiction approaches authorship identically.
South Korea: Its copyright framework combines specified exceptions with a broader fair-use provision. This makes it a useful comparison for examining how defined permissions and flexible standards can operate within the same system.
China: Its statutory framework approaches permitted uses differently from the United States’ open-ended fair-use model. The comparison underscores why assumptions about permissible AI-related uses cannot simply be transferred from one country to another.
These differences affect more than legal terminology. They influence how creators assess their rights, how developers evaluate risk, and how businesses plan training and deployment.
The question is therefore not just whether copyright law can address AI. It is how differently it does so across borders.
When Legal Differences Become a Business Strategy:
One concern explored in the paper is regulatory arbitrage: the possibility that organizations structure activities to take advantage of differences between national legal systems.
For AI development, those differences could influence where training occurs, how datasets are obtained, or which licensing arrangements are pursued. A company may consider one jurisdiction more favorable because of its exceptions, enforcement environment, or other legal conditions.
That does not mean moving an operation makes every use lawful. Cross-border activities can still implicate rights and obligations elsewhere. Nor is choosing a favorable jurisdiction necessarily unlawful in itself.
The concern is that fragmented rules can make regulatory advantage a significant factor in technological development. Creators whose work circulates globally may then face a difficult task: understanding and enforcing rights across systems that treat comparable activities differently.
Developers also face uncertainty. A practice that appears permissible in one jurisdiction may encounter restrictions in another.
Why Human Creative Agency Still Matters:
Copyright protects more than ownership of a finished product. Economic rights help rights holders control and benefit from certain uses of protected works. Moral rights, where recognized and applicable, protect interests such as attribution and the integrity of a work.
AI raises questions about both dimensions. Who receives recognition for creative contributions? Who can authorize relevant uses? How can creators exercise their rights when the use of their work is difficult to identify?
Human creative contribution remains central to these questions, even though the jurisdictions examined do not apply identical authorship rules.
This does not require treating AI as incompatible with creativity. AI can be a tool within a human creative process. The challenge is to distinguish that use from situations in which human contributions, or the rights attached to existing works, become obscured.
What a More Workable Approach Could Look Like:
The paper argues for greater international alignment centered on protecting human authorship. Its recommendations focus on making rights more practical to understand and exercise.
Transparency and attribution are an important starting point. Appropriate disclosure obligations could help rights holders identify relevant uses of their work and support accountability. But disclosure alone is not permission, and attribution alone does not resolve whether compensation or authorization is required.
Licensing and remuneration mechanisms also need to address the scale of AI development. Where authorization is required, workable licensing arrangements, including collective management where appropriate, could help connect lawful access with compensation.
Technical standards can support those mechanisms. Rights information and other technical tools are more useful when connected to legal duties and enforceable processes, rather than treated solely as voluntary commitments.
Finally, international coordination could help reduce uncertainty without requiring every country to adopt identical laws. WIPO can support intellectual-property dialogue and cooperation, while the WTO’s relevance lies in trade-related intellectual-property obligations, particularly the TRIPS Agreement. Their roles should be understood within those distinct mandates.
The Bigger Question:
The future of AI and copyright is not only about whether machines can generate convincing creative content. It is about how societies recognize and protect the people whose expression contributes to that technological development.
Fragmented national approaches leave creators and rights holders navigating inconsistent rules. They also leave developers without clear expectations across markets.
A more coordinated framework could make room for innovation while keeping human creative agency, lawful use, and meaningful protection at the center of the discussion.
Read the full research paper below for the comparative analysis of the United States, the United Kingdom, South Korea, and China, and some ideas for a more coordinated international approach.
Comments