Can OpenAI assert its own originality while drawing heavily from Ghibli’s iconic style, or is it simply algorithmic finesse cloaked in another’s legacy?
Ghibli as a Display, Not a Trait
In a time when artificial intelligence is already capable of writing, speaking, and even illustrating like humans, OpenAI has pushed the envelope by teaching it to envision.
On March 25, OpenAI unveiled a new feature in its GPT-4o model that swiftly captured the attention of the internet. Known as “4o Image Generation,” this tool allows users to create breathtaking, photorealistic images using only text prompts.
What sparked the most excitement online, though, was the AI’s remarkable ability to emulate the cherished animation style of Studio Ghibli.
Studio Ghibli, the legendary Japanese animation studio, is celebrated for its handcrafted aesthetic and emotionally charged narratives. Its unique visual language is so identifiable that even the smallest attempt to imitate it is instantly recognized.
Consequently, when users on ChatGPT Plus, Pro, and Team tiers began posting AI-generated images in the beloved Ghibli style, reactions varied from awe to skepticism. The visuals were so convincing that many viewers assumed they were created by artists rather than an algorithm.
OpenAI’s CEO appeared to fully embrace this viral moment, switching his profile photo on X to a Ghibli-inspired image of himself and encouraging users to create even better renditions.
According to OpenAI, the 4o image tool stands as its most sophisticated visual generator yet, crafted to yield content with “precision, accuracy, and photorealism.” What sets it apart from previous iterations is its embedded multimodal intelligence.
This innovation enables it to not only transform text into imagery but also comprehend context, emotional nuances, and artistic indicators. Essentially, it interprets the material similarly to how an illustrator might discern the subtleties of a narrative.
However, such powerful tools also come with their challenges. The unexpected surge in use strained OpenAI’s server capacity to its limits.
Altman admitted that the company’s GPU infrastructure was “melting” under the pressure, prompting OpenAI to temporarily limit access for free-tier users and implement usage caps on image generation to alleviate the burden.
For the time being, users are delighting in their AI-generated visual stories. Yet, beneath this enthusiasm, deeper inquiries are starting to emerge.
The Artistic Controversy: Imitation or Innovation?
While social media buzzed with enchanting Ghibli-style portraits created by ChatGPT’s new image generator, a contrasting response was unfolding in other online spaces.
The most significant pushback came from artists and creatives who perceived a major tech corporation profiting from what seemed to be artistic imitation — devoid of credit, consent, or recompense.
On X, some users criticized the AI-generated artwork as hollow reproductions lacking the “emotion, depth, and soul” that animators infuse into every frame of Ghibli films.
Some posts went further, accusing OpenAI of “plagiarizing” decades of handcrafted artistry and storytelling. A commenter described it as “identity theft in the history of art,” highlighting the striking resemblance of the Ghibli-style outputs.
Others expressed their dissent more bluntly, dubbing the new feature as “a plagiarism machine” and accusing the company of “stealing Studio Ghibli’s artistry.” One user provocatively asked, “Would you like it if I took your designs and never compensated you?”
Karla Ortiz, a professional illustrator known for her collaborations with Marvel and Dungeons & Dragons — and one of the artists currently litigating several AI companies for utilizing her copyrighted works — also chimed in.
She remarked that OpenAI’s Ghibli-style functionality exemplifies how AI companies “neglect the efforts of artists,” labeling the image generation as a form of exploitation.
Ortiz’s ongoing lawsuits are part of a larger effort to hold AI developers responsible for training their models on extensive datasets that incorporate copyrighted works, often gathered without permission or transparency.
The backlash also resonated with a warning previously articulated by Studio Ghibli’s co-founder, Hayao Miyazaki. In a well-circulated 2016 interview, after watching a snippet of early AI-generated animation, he voiced his concerns candidly.
“I can’t watch this and find it interesting. Whoever creates this kind of content clearly has no understanding of what real pain is. I’m utterly disgusted. If you truly want to make creepy things, go ahead — but I would never want to incorporate this technology into my work. I strongly feel that it’s an insult to life itself,” he declared.
What Laws May Miss, AI Can Appropriately Claim
As public discontent mounted, many started calling on Studio Ghibli to consider legal action. However, there’s a significant hurdle: under current legislation, the studio may find itself with limited options — particularly in Japan, where it operates.
Japan, unlike most major economies, has adopted a notably flexible approach to AI and copyright. A report outlines Japan as the only prominent nation where AI models can legally train on copyrighted content without requiring permission from the original creators.
This means that even if OpenAI had used frames from acclaimed films like Spirited Away or Princess Mononoke to train its image generator, it wouldn’t necessarily breach Japanese laws.
This legal ambiguity places Ghibli — and other artists potentially affected — in a precarious situation. The law has not kept pace with the rapid development of AI, creating a gray area that companies can exploit.
Ironically, just a day before this controversy gained momentum, The New York Times secured the right to move forward with its lawsuit against OpenAI regarding the extensive use of its written works to train ChatGPT.
Although this case pertains to text rather than visuals, it raises parallel questions: Can AI learn from copyrighted content without consent — and then produce works that closely resemble it?
OpenAI asserts it has implemented precautionary measures. In a recent paper, the company noted that it has established a “refusal” mechanism to prevent image generation in styles of living artists.
Nevertheless, these protections do not appear to apply to brands or deceased creators. Since Studio Ghibli operates as a studio rather than an individual, its identifiable visual style seems to escape those restrictions.
This is where the legal murkiness deepens. While trademarks can safeguard logos, characters, and specific images, the protection of artistic style remains elusive under present copyright laws.
One may successfully trademark Totoro. But one cannot trademark “whimsical, hand-drawn landscapes imbued with a sense of magical realism and emotional tranquility” — even if that essence encapsulates Ghibli’s unique essence. Currently, there is no global legal standard for safeguarding artistic style in isolation.
And this situation likely won’t be the last instance of its kind. As AI technology advances, future models are bound to create music, edit films, and replicate entire creative aesthetics — often with alarming precision.
When that time comes, society will have to confront more substantial questions: Do we favor authenticity or expediency? Emotional depth or boundless output?
Ultimately, this discussion transcends Ghibli. It delves into the significance of the human element in creativity — and whether that distinction holds when machines can replicate it indistinguishably.