Ownership Disputes Over AI-Generated Content In Corporate Environments.
1. GEMA v. OpenAI (Germany, 2025)
Issue: Whether training AI on copyrighted content and reproducing it constitutes infringement.
Facts:
GEMA, the German music rights organization, sued OpenAI because its AI models had been trained on copyrighted song lyrics.
The AI could reproduce lyrics when prompted.
Court’s Decision:
The Munich Regional Court ruled that reproducing copyrighted lyrics from memory is unauthorized reproduction.
Text and data mining exceptions did not apply because the AI was reproducing complete works, not just analyzing them.
OpenAI was ordered to stop distributing infringing lyrics and to pay damages.
Significance:
Shows that even training AI on copyrighted works can create liability.
Establishes that corporations operating AI systems may be responsible for outputs, not just end-users.
2. Ultraman AI Copyright Case (China, 2024)
Issue: AI-generated images resembling copyrighted works.
Facts:
A Chinese company used an AI generator to create images similar to the Ultraman franchise.
The copyright holder sued, claiming these were unauthorized reproductions.
Court’s Decision:
The Guangzhou Internet Court held that the AI-generated images were substantially similar to protected works and were derivative works.
AI output was considered infringing even though it was generated by a machine.
Significance:
Demonstrates that courts can treat AI-generated outputs as copyrightable derivative works.
Training and generating similar works without permission may constitute infringement.
3. Andersen, McKernan & Ortiz v. Stability AI / Midjourney (U.S., ongoing)
Issue: Liability of AI image generators for using copyrighted images in training.
Facts:
Artists sued Stability AI and Midjourney for training AI models on copyrighted images scraped from the internet.
AI outputs allegedly replicated or resembled the plaintiffs’ works.
Developments:
Courts are considering whether training constitutes copyright infringement or qualifies as fair use.
Major studios have filed similar suits for unlicensed use of iconic characters.
Significance:
Highlights the tension between fair use defenses and the rights of copyright owners.
Raises corporate risks for AI developers using large datasets of copyrighted material.
4. Zarya of the Dawn (U.S., 2023)
Issue: Can AI-generated artwork be copyrighted?
Facts:
A comic book fully illustrated by AI was initially registered for copyright.
It was later discovered that AI created all illustrations with minimal human intervention.
Court/Agency Decision:
The U.S. Copyright Office revoked copyright protection for AI-created images.
Only the human-authored text/story remained copyrightable.
Significance:
Reinforces that pure AI output is not protected under U.S. law.
Corporations must establish human creative contribution for AI outputs to be owned and protected.
5. Thaler v. U.S. Copyright Office (U.S., 2020s)
Issue: Can AI itself be recognized as the author of a copyrighted work?
Facts:
Stephen Thaler tried to register AI-generated art created by his “Creativity Machine.”
Decision:
The U.S. Copyright Office denied registration, reaffirming that human authorship is required.
Significance:
AI cannot legally own copyright.
Ownership claims must rest on human creators or contractual arrangements.
6. Entrepreneur Media LLC v. Meta Platforms Inc (U.S., 2025)
Issue: AI training on copyrighted texts without permission.
Facts:
Entrepreneur Media sued Meta for using its copyrighted business books to train AI models.
Legal Issue:
Meta claimed fair use, arguing the training was transformative.
The case explores the boundaries of copyright in AI training datasets.
Significance:
Corporations must be cautious using copyrighted content to train AI.
Fair use arguments are context-dependent and not guaranteed.
7. Disney / Universal / Warner Bros. v. MiniMax AI (U.S./China, 2025)
Issue: AI platforms generating unlicensed content based on iconic characters.
Facts:
Major studios sued AI companies for generating images and content using copyrighted characters without permission.
Significance:
Shows that even well-known franchises face corporate enforcement issues when AI outputs mimic their works.
Corporations using AI must verify that content doesn’t infringe third-party rights.
Key Legal Principles Emerging from These Cases
| Principle | Implication for Corporations |
|---|---|
| Human authorship is required | Pure AI output is generally not copyrightable; human involvement is essential. |
| Training on copyrighted material | Can create liability (Germany, China), fair use may apply in U.S., but context-dependent. |
| AI-generated derivatives | Outputs replicating copyrighted works can infringe even without human copying. |
| Corporate responsibility | Companies may be liable for AI outputs, so internal policies are crucial. |
| Contractual assignment | Corporations must define ownership of employee-generated AI content. |
Practical Takeaways for Corporations
Define ownership clearly in employment contracts or policies.
Monitor AI training datasets for copyrighted materials.
Ensure human creative contribution to claim ownership.
Check jurisdictional differences — U.S., EU, and China handle AI copyright differently.
Consider licenses for copyrighted works used in AI training.
RELATED Blog
comments