Who Owns the Copyright of AI-Generated Designs? (A Comparative Analysis of U.S., European, and Korean Approaches)

 

Still confused about copyright for AI-generated designs? Me too. Here’s a deep dive into the subtle differences in the legal approaches of the U.S., Europe, and Korea, based on the latest case law and legislation, along with a practical guide for creators.

AI Copyright: A Deep Analysis of Case Law and Legislation! This article breaks down the complex copyright issues surrounding designs created with generative AI like Midjourney and ChatGPT. Learn how to protect your creative work with our checklist and review of the latest legal challenges in different countries.

 

Hey there! In an era where we can whip up stunning designs with AI, a critical question has captured everyone’s attention: “Who actually owns the incredible designs made by AI?”

While the legal frameworks for AI-assisted designs in Europe, the U.S., and Korea are all fundamentally based on the ‘human author’ principle, they each show subtle differences in focus due to their unique legal traditions and policy approaches. Today, we’ll take a closer look at these key distinctions through the lens of the latest case law and legislation.

 

🌍 Key Differences in AI Copyright Law Across the Globe

🇪🇺 European Union: Focus on the Designer’s Creative Choices

The core question in the European Union is this: “Does the design reflect the designer’s own personality and creative choices, rather than just its aesthetic effect or novelty compared to prior works?” And, “Was there room for ‘free and creative choices’ in the expression of the work, ensuring it wasn’t solely dictated by its technical function?” In other words, the emphasis is placed more on the creative process than on the final outcome.

This standard, known as the ‘Author’s Own Intellectual Creation’ (AOIC), has been established through various court rulings.

• The Ganni Shoe Case (2024): This case showed that even if a designer is inspired by existing designs, they can still meet the AOIC standard by making creative choices that reflect their own personality.
• The AI Act (2024): This law requires AI providers to disclose summaries of their training data, making it easier to establish a causal link in copyright infringement disputes.

A New Proposal to Watch: The ‘Recognizability Test’
Recently, Advocate General Szpunar of the Court of Justice of the European Union (CJEU) proposed a new ‘Recognizability Test,’ suggesting that infringement occurs if the creative elements of the original work are ‘recognizable’ in the copy. However, this proposal has faced criticism for potentially stifling innovation in industries like fashion and unfairly increasing legal liability for designers. Its future is one to watch.

🇺🇸 United States: A Strict Scrutiny of ‘Sufficient Human Intervention’

The central question in U.S. law is: “Does the expressed design show a sufficient degree of intervention or modification by a human author to be recognized?” The U.S. applies the ‘human author’ principle very strictly, making it clear that works generated solely by AI cannot be copyrighted.

• The Stephen Thaler Ruling (2023): The court denied copyright for a work created solely by AI, reaffirming the ‘human author’ requirement.
• ‘Transformative Use’: This legal concept provides a path to copyright protection if a human’s modifications to an AI-generated work are substantial enough to create a new expression, meaning, or message.
• The Kristina Kashtanova Case (2022): While the graphic novel as a whole—with its creative ‘selection and arrangement’ by a human—was granted copyright, the individual AI-generated images were not protected, highlighting the importance of ‘sufficient intervention.’

🇰🇷 South Korea: Based on ‘Additional Work’ and ‘Creative Arrangement’

In Korean law, the key questions are: “Did a human perform ‘additional work’ such as modification or enhancement on the AI output?” or “Is there creativity in the way the AI outputs were selected and arranged?” The ‘Generative AI Copyright Guide’ explains that while AI outputs themselves are not protected, copyright can be recognized for the parts that involved human ‘additional work.’ Furthermore, if there is creativity in the selection and arrangement of AI outputs, the work may be recognized as a copyrightable ‘compilation.’

While the Ministry of Culture, Sports and Tourism and the Korea Copyright Commission have been leading the charge on these issues, it remains to be seen which government body will take the lead following the establishment of the new Ministry of Intellectual Property in 2025.

 

📊 A Side-by-Side Comparison of Legal Approaches

Category	European Union (EU)	United States (US)	United Kingdom (UK)	South Korea
Copyright Requirements	- AOIC (Author’s Own Intellectual Creation) - The author’s unique personality and creative choices - Freedom of choice not dictated by technical function	- Strict human authorship principle - No protection for purely AI-generated works - AI-assisted works protectable with sufficient human input	- Gradual adoption of AOIC standard - Emphasis on the designer’s personality and free, creative choices	- Expression of human thoughts or feelings - No protection for AI outputs themselves - Copyright recognized only for the ‘additional work’ by a human
Infringement Criteria	- Restricted act, causal link, and substantial similarity - Replication via TDM is considered copying - Controversial ‘Recognizability Test’ proposed by AG Szpunar	- Transformative Use - Replication via TDM is copying, but fair use exceptions apply - Adds new expression, meaning, message, or utility	- Restricted act, causal link, and substantial similarity - Replication via TDM is considered copying - Replicating ‘style’ alone is not infringement	- Restricted act, causal link, and substantial similarity - Replication via TDM is considered copying - Replicating ‘style’ alone is not infringement
Policy & Regulation	- AI Act: Mandates disclosure of training data summaries - Reports and policy papers from the European Parliament	- Executive orders and agreements on safety measures like watermarking - Leading international discussions through WIPO	- No specific legislation yet on AI and IP - Reaffirmation of the AOIC standard’s validity	- AI-Copyright Working Group launched in 2023, published guidelines - Recommends documenting process and data sources - Future changes expected with new Ministry of Intellectual Property in 2025

 

🎯 A Practical Checklist for Creators Using AI

✅ How to Increase the Likelihood of Copyright Protection

1. Lead with a Human-Driven Plan: Create a detailed brief with your concept, style, and color palette. Experiment with multiple prompts and curate the results. (This helps prove ‘creative choices’ in the EU and ‘creative arrangement’ in Korea).
2. Make Significant Edits: Use tools like Photoshop or Illustrator to fine-tune AI outputs. Add your own text and graphic elements to make the work truly yours. (This helps demonstrate ‘sufficient intervention’ in the U.S. and ‘additional work’ in Korea).
3. Document Your Process Thoroughly: Keep a record of your prompt iterations, editing steps, and the reasoning behind your final decisions. This documentation can be crucial evidence of your creative contribution. (Important in all jurisdictions).

❌ When Copyright Protection is Unlikely

• Using the direct output from a simple prompt with no changes.
• Simply copying and pasting various AI-generated results together without creative arrangement.
• Having no records or proof of your creative process.

 

🚀 The Future for AI and Designers

Looking ahead, we can expect copyright guidelines for AI to become more specific and for detection technologies like watermarking to improve. We’ll also likely see more international cooperation to harmonize these standards.

💡 Practical Tips for Designers

1. Meticulously Document Your Creative Process: Save screenshots, screen recordings, and notes on your prompt development to create a paper trail proving your ‘human contribution.’
2. Use AI as a Tool, Not a Replacement: Leverage AI for brainstorming or creating initial drafts, but always infuse the final product with your own creativity.
3. Be Transparent: Be open with clients and the public about how you use AI tools to build trust and manage expectations.
4. Stay Informed: Keep an eye on legal changes, court rulings, and industry guidelines in your country to minimize your legal risks.

Key Takeaways on AI Design Copyright

Core Principle: All legal frameworks require ‘human creative intervention.’

Key Differences: The U.S. tends to focus on whether the result has been “sufficiently modified,” Europe emphasizes the “creative choices” made in the process, while Korea places weight on the “additional work” applied to the outcome.

Best Practice:

When using AI, document your entire process from planning to final edits!

Future Tech: AI watermarking is emerging as a key technology for content transparency.

Use these tips to create smarter with AI!

🎬 Final Thoughts: Becoming a Wise Creator in the AI Era

AI has become a powerful partner in our creative endeavors. But it’s crucial to remember that technology is just a tool—the true author is still human.

The common thread across all countries is the emphasis on ‘human creative intervention.’ Infuse your unique ideas and sensibilities into AI-generated outputs. That’s what transforms a machine’s product into a genuinely creative work that can be protected by copyright.

AI technology will continue to evolve, and the laws governing it will become more refined. But one thing won’t change: a creator’s imagination and expression are their most valuable assets. Remember, AI isn’t here to replace your creativity—it’s here to amplify it! ✨

 

Frequently Asked Questions

Q: If I edit an AI image in the U.S., is it automatically copyrighted?

A: Not necessarily. U.S. courts look for a significant level of creative modification that adds new expression or meaning—enough to qualify as ‘transformative use.’ Simple changes like adjusting colors or cropping might not be enough.

Q: What does the ‘AOIC’ standard in Europe really mean?

A: It stands for ‘Author’s Own Intellectual Creation’ and focuses on whether the author was able to make ‘free and creative choices.’ This means the final work can’t be something automatically determined by technical constraints; it must reflect the author’s personal touch and choices to be copyrightable.

Q: How can I get an AI output recognized as a ‘compilation’ in South Korea?

A: You need to do more than just collect AI-generated outputs. There must be creativity in the ‘selection, arrangement, or composition’ of the materials. For example, if you curate specific AI images to tell a unique story and arrange them in a sequence that conveys a new meaning or feeling, you have a stronger case for a copyrightable compilation.

I hope this article was helpful for all you creators navigating the age of AI. If you have any questions, feel free to ask in the comments! 😊

※ Legal Notice ※
The information provided in this blog post is for general informational purposes only and is not a substitute for professional legal advice. Please consult a qualified attorney for advice on your specific legal issues.