Ownership Disputes In AI-Assisted Screenplay Writing And Narrative Generation
📌 1. Zarya of the Dawn — AI-Assisted Comics and Copyright Limits (US)
Jurisdiction: United States
Core Issue: Can AI-generated art be copyrighted, and what parts of a mixed AI-human work are protected?
Facts
Zarya of the Dawn is a comic written by Kris Kashtanova with images generated by Midjourney (an AI image generator).
Legal Dispute
Kashtanova initially obtained copyright protection for the work. However, after the U.S. Copyright Office learned that the artwork was produced by AI, it revoked the copyright protection for those illustrations, on the reasoning that copyright requires human authorship, and AI is not recognized as a human author under U.S. law.
Outcome & Legal Significance
The text and arrangement (selection and organization) of the comic were recognized as copyrightable because they reflect human creativity.
The AI-generated artwork, however, was not protected because it lacked sufficient human authorship — confirming that in the U.S., only works with human creative input receive copyright.
👉 Takeaway: Works involving AI may be partially protected, but courts and copyright offices will strip protection from elements that are the product of AI alone without substantial, identifiable human authorship.
📌 2. Thaler v. U.S. Copyright Office / DABUS — Federal Appeals Court Rejects AI Authorship (US)
Jurisdiction: United States
Core Issue: Can a non-human (AI) be an author for copyright purposes?
Facts
Stephen Thaler sought copyright protection for an artwork his AI system (called DABUS) allegedly created without human intervention.
Legal Dispute
Thaler argued the AI should count as the author and that the artwork deserved copyright protection.
The U.S. Copyright Office denied the application, and the case went to the federal appeals court.
Decision
The U.S. Court of Appeals for the D.C. Circuit agreed with the Copyright Office, holding that:
Copyright law requires human authorship.
AI alone — without human creative choice — cannot be the author of a copyrighted work.
Outcome & Legal Significance
Thaler’s artwork remains unprotected under copyright. This case reinforces a clear line: machines do not have legal personality or authorship rights under current U.S. copyright law.
👉 It’s one of the most definitive rulings in the U.S. establishing the human authorship requirement for AI-generated creative works.
📌 3. Artists’ Lawsuit Against AI Companies (Stable Diffusion / Midjourney) — Copyright Infringement (US)
Jurisdiction: United States
Core Issue: Do AI generators infringe on artists’ rights by using copyrighted works as training data?
Facts
Three visual artists — Sarah Andersen, Kelly McKernan, and Karla Ortiz — sued Stability AI, Midjourney, and DeviantArt for allegedly training their AI image generators on billions of images without permission.
Legal Dispute
Plaintiffs claim their copyrighted art was used to train these AI systems.
They argue that AI output reproducing their style or substantially similar content amounts to unauthorized derivative works or infringement.
Court Activity
A federal judge initially questioned the accuracy of some claims but allowed the artists to reframe their complaint.
Outcome & Legal Importance
This case is ongoing, but its significance lies in:
Challenging the idea that training a model using copyrighted art is automatically lawful.
Pushing courts to define whether AI outputs are derivative of training material and whether training itself constitutes infringement.
👉 This dispute illustrates how ownership issues spill into data rights and training processes, not just final outputs.
📌 4. Anthropic & Meta AI Copyright Cases — Fair Use Defenses (US)
Jurisdiction: United States
Core Issue: Does training AI on copyrighted materials constitute infringement or is it fair use?
Cases Argued
a) Writers vs Anthropic (Claude AI)
A class action by journalists claimed Anthropic used pirated books for training.
Court found the use transformative enough for fair use, but allowed infringement issues to move forward on damages.
b) Writers vs Meta (LLaMA)
Authors including Sarah Silverman sued Meta for training LLaMA on their books.
The court ruled plaintiffs failed to show market harm, granting Meta a favorable ruling but not closing the door on compensation claims.
Outcome & Legal Importance
These cases demonstrate the fiery debate over whether training AI on copyrighted material is lawful.
Courts weigh whether the AI’s use is transformative and whether it harms the market for the original works.
👉 Ownership disputes include data rights and training practices, not just output rights.
📌 5. Li v. Liu — Human-AI Collaboration Recognized by Chinese Courts (China)
Jurisdiction: China
Core Issue: When can AI-assisted creative outputs be copyrighted if there’s human input?
Facts
In Li v. Liu, the Beijing Internet Court considered whether images generated by AI — where humans crafted detailed prompts and refined outcomes — were sufficiently creative.
Decision
The court found that:
The human’s original input, choices, and refinements contributed creative expression.
The resulting AI-assisted work satisfied criteria for originality and thus could be protected.
Outcome & Legal Importance
This case contrasts sharply with U.S. rulings:
It recognizes that significant human intervention and prompt design can confer copyright.
It provides a model for courts outside the U.S. to balance human and AI contributions in creative works.
👉 This illustrates differing global approaches to AI authorship and ownership.
📌 6. Related Indian Examples: AI Copyright Applications & Human Authorship (Supporting Context)
Legal Context: India
Relevant Administrative Rulings
The Indian Copyright Office in a famous “Raghav AI” case refused to assign copyright to an artwork solely invented by AI, because AI cannot be an author under Indian law.
Key Principles Recognized
Human involvement is required for copyright protection.
AI may be listed in filings but lacks legal authorship status.
👉 This aligns broadly with U.S. positions and shows parallels in Indian jurisprudence and copyright practice.
📌 Key Legal Themes Across These Cases
📍 1. Human Authorship Is Often Essential
Most jurisdictions treat creative input by a human as necessary to qualify for copyright — AI alone cannot be “author.”
Zarya of the Dawn revoked AI imagery protection.
Thaler / DABUS confirmed no copyright for purely AI works.
📍 2. AI as a Tool — Not a Creator
If AI is used as a tool in the creative process, courts focus on who exercised creative judgment (prompts, edits, sequencing).
China’s Li v. Liu found copyright due to human creative contributions.
📍 3. Disputes Extend to Training Data
Artists and authors are suing AI companies, claiming unauthorized copying and derivative works from training processes.
The Andersen v. Stable Diffusion / Midjourney litigation.
Writers vs Anthropic / Meta AI cases.
📍 4. Fair Use Is Central But Not Settled
Courts are still balancing:
Whether training data use is fair use/fair dealing.
Market harm and transformation factors.
📍 5. Jurisdictional Variation Is Real
The U.S. and China take different stances on AI-assisted works, showing that ownership disputes have global but not uniform answers.
📌 Conclusion — Screenplays & Narrative AI
When it comes to AI-assisted screenplays and narrative generation, courts are grappling with:
Who owns the output? Only humans typically hold the copyright unless their role is trivial.
Is AI content protectable? Pure AI generated content often lacks protection.
Does training on copyrighted works constitute infringement? Some courts allow challenges to proceed; others find fair use.
Practical Implication for Creators and Studios:
To claim ownership, a human must meaningfully guide the AI (prompts, revisions, editing).
Contracts should clearly assign rights and define human contribution.
Be prepared for litigation over both output and training data rights.
RELATED Blog
comments