Defending Your Art From AI Mimicry

Authored by Babak Akhlaghi on April 4, 2025. When OpenAI released an image generation feature capable of producing images that look remarkably like Studio Ghibli animations, the internet exploded with “Ghiblified” pictures. The images went viral, prompting OpenAI to quietly pull back the feature from free-tier rollout. This controversy highlights a fundamental question I’ve been wrestling with as both a legal professional and an observer of creative industries: Who owns the copyright to AI-generated art that mimics a distinctive artistic style?

The answer isn’t simple. To understand why, we need to examine the foundations of copyright law itself.

Copyright Basics Weren’t Built for AI

The Copyright Act protects original works of authorship fixed in any tangible medium of expression. Copyright protects the specific expression of an idea, not the underlying ideas or techniques themselves. When you upload your photo to generate a “Ghiblified” image, you probably have a copyright claim to your original photo. But who owns the copyright to the AI output?

This question forces us to confront a fundamental limitation of our current legal framework: copyright protects specific expressions, not general styles or approaches to creating art. An artist’s style is generally considered a method or approach to creating art, falling under the category of ideas or techniques that copyright explicitly excludes from protection (17 U.S.C. § 102(b)).

But the waters get murkier when we consider how AI systems learn to mimic artistic styles. They’re trained on datasets of images—often without the original artists’ permission.

The Thomson Reuters Precedent Matters

A recent case provides crucial insight into how courts might view AI training on copyrighted material. In Thomson Reuters Enter. Ctr. GmbH v. Ross Intel. Inc., No. 1:20-cv-613-SB (D.Del. Feb. 11, 2025), the U.S. District Court for Delaware rejected a fair use defense in the context of AI training data. This creates a significant precedent.

Ross Intelligence had used Thomson Reuters’ legal headnotes to train its AI legal research tool without authorization. The court ruled against fair use because Ross’s AI essentially served the same purpose as Westlaw, owned by Thomson Reuters—providing legal research results based on training on Thomson Reuters data.

This case suggests that if an AI system is trained on Studio Ghibli works without authorization and produces outputs mimicking their style, it might constitute copyright infringement.

Fair Use in the AI Context

The fair use doctrine, codified under 17 U.S.C § 107, provides an exception to copyright infringement for certain uses of copyrighted material. Courts consider four factors:

1. The purpose and character of the use (commercial vs. educational)

2. The nature of the copyrighted work

3. The amount and substantiality of the portion used

4. The effect of the use upon the potential market for the original work

One critical issue that needs clarification is what constitutes “transformative use” in an AI context. The Thomson Reuters case wrestled with this question. The court noted that “if an original work and a secondary use share the same or highly similar purposes, and the second use is of a commercial nature, the first factor is likely to weigh against fair use.”

What makes the Thomson Reuters case particularly interesting, however, is that Ross didn’t display the headnotes to customers. Instead, the copying occurred during an intermediate step, where Ross converted the headnotes into numerical data to train its AI. The court distinguished this from previous cases where intermediate copying of computer code was deemed fair use because those cases involved a “necessity for innovation” that wasn’t present in Ross’s case.

An important distinction worth emphasizing: the Thomson Reuters case involved non-generative AI. The court explicitly limited its ruling to that context. With generative AI, like the system producing Studio Ghibli-style images, the analysis might differ—but would still depend on whether the AI output significantly alters the purpose and meaning of the original work.

Immediate Protection Strategies for Artists

If we accept that current copyright frameworks inadequately protect artistic style in the AI era, what can artists do now?

The next best protection is through contractual agreements. Artists should carefully read terms of service before uploading their work to AI platforms. Fine print in these agreements may enable AI systems to use their work for training and generate similar style outputs.

Litigation is expensive and often beyond the reach of individual artists, especially when going against corporations with vast resources. I envision organizations emerging specifically to protect creators against AI companies. These could be membership-based groups that pursue copyright infringement claims when member artists’ rights are violated, ensuring proper compensation and leveling the playing field against companies with deeper pockets.

The Ethical Dimension Beyond Legal Requirements

There should be ethical boundaries that AI technology companies respect regardless of legal requirements. Without this, we risk losing the human touch that makes art meaningful.

Currently, AI systems don’t experience emotions—they can’t feel pain, suffering, or joy. Their ability to capture these elements in art comes purely from training on human-created data. As AI systems evolve toward more general intelligence, they may better understand or even experience emotions, making their generations increasingly realistic and closer to masters like Miyazaki-san. This prospect is both fascinating and frightening.

The race for AI supremacy complicates these ethical considerations. OpenAI’s CEO has asked the U.S. government to allow AI models to train on copyrighted material to strengthen America’s lead in AI technology. But this approach seems shortsighted. If all copyrighted material becomes fair game for AI training, what happens when these systems consume everything available? Artists may stop creating or publishing work, knowing it will be replicated by AI without repercussions.

The race to AI supremacy should not come at the expense of creators. It will eventually backfire. Yet if other countries allow their AI systems to train on these materials while countries with stronger standards restrict this practice, the latter may fall behind technologically. Perhaps we need international laws governing AI use that all countries should follow.

Balancing Innovation and Protection

I don’t believe a perfect solution exists. We should strive for balance, with policies consistent with the fundamental purpose of our intellectual property system: promoting the progress of science and useful arts.

Economic models where artists receive compensation when AI systems use their work for training represent one potential solution. This approach acknowledges both the value of the original creative work and the potential for AI to build upon it in ways that further human creativity.

Distinguishing Between AI Systems

Existing law can address distinctions between different types of AI systems. In a generative AI context, where the system creates entirely new content, the third fair use factor (amount and substantiality of the portion used) might favor the AI technology company.

However, this is just one of four factors. If we apply all four to the Studio Ghibli controversy:

1. If the AI output’s purpose is to mimic Studio Ghibli’s style for commercial purposes, this factor likely favors Studio Ghibli.

2. Since Studio Ghibli’s works are highly creative, this factor also favors Studio Ghibli.

3. If the generated AI output is completely new, this factor might favor the AI company.

4. If AI outputs would draw customers away from Studio Ghibli’s market, this factor would likely favor Studio Ghibli.

Balancing these factors requires nuanced judgment, but our current legal framework provides the tools for this analysis.

Protecting Distinctive Style

I don’t believe we need an entirely new category of intellectual property to protect artistic style. While copyright doesn’t protect style itself, an artist might use trademark law to protect a distinctive brand or signature associated with their work. If an artist’s style is closely tied to their brand identity—like a specific way they sign their work or distinctive elements that identify the source—trademark protection might apply.

International Challenges

Most advanced countries are signatories to the Berne Convention, which establishes a consistent baseline for copyright protection. However, enforcement varies significantly. Countries like China have historically had less stringent IP enforcement than the United States.

This enforcement disparity drives arguments from companies like OpenAI that U.S. government should loosen restrictions on training with copyrighted material to remain competitive. The problem with this argument is that it undermines the fundamental purpose of intellectual property protection—encouraging creation. If AI companies can freely appropriate on creative works without compensation, artists lose motivation to create, and the global creative industry will suffer in the long run.

Looking Forward

Despite these challenges, I remain optimistic about the future relationship between human creators and AI tools. When artists harness AI as a creative tool rather than seeing it as a replacement, it may elevate human creativity to new heights that benefit us all.

For now, my practical advice to artists concerned about AI systems mimicking their style is straightforward: don’t upload your creative work to these systems, as you may inadvertently grant them rights to train on it. When selling your work, attach contractual provisions that limit what buyers can do with purchased items—specifically prohibiting them from uploading the work to AI systems or sharing it online in ways that make it available for AI training.

The legal landscape for AI and artistic expression remains in flux. While courts and legislators catch up to technological reality, artists must be proactive in protecting their creative identities. The goal isn’t to halt technological progress but to ensure that human creativity remains valued and protected in an increasingly AI-driven world.

Stay Informed with Expert Insights!

Want more valuable tips on navigating patent law and protecting your intellectual property? Join our free weekly newsletter for the latest updates, expert advice, and exclusive content straight to your inbox. Subscribe today and never miss out on essential information that can make all the difference in your patent journey!  Subscribe to our Newsletter