The 2024 film Hayao Miyazaki and the Heron tells the story of Miyazaki, the founder of the animation studio Studio Ghibli, and documents how his painstaking work consumes him; his film The Boy and The Heron took seven years to make.
It’s worth bearing this in mind as we recall how last week, following ChatGPT’s 4o image generation update, it was let loose on the world and a trend quickly emerged. Users began creating images in the style of Studio Ghibli’s animated characters.
The results were mainly people messing around and creating a character in their own image or famous people in meme-able situations, which, while on the surface seemed harmless enough, also drew criticism from those explaining that Studio Ghibli’s universe is carefully curated and draws on the natural world. The irony of this was also raised, given the energy resources that AI draws on.
Advertisement
Then there were the rather dark and surprising uses, such as the US Government knocking this up in poor taste.
https://t.co/PVdINmsHXs pic.twitter.com/Bw5YUCI2xL
— The White House (@WhiteHouse) March 27, 2025
The fallout was fast and criticism strong, propelled in particular by a resurfaced comment from Miyazaki, recorded in the 2016 documentary Never-Ending Man: Hayao Miyazaki, where he presciently says: “I am utterly disgusted… I strongly feel that this is an insult to life itself” when shown an example of AI’s capabilities.
Granted, this was some time ago and AI has moved on some way, but it’s pretty unlikely Miyazaki will have been clapping and laughing along with the trend that is essentially a social media filter.
With OpenAI’s resources and reach, ChatGPT’s brand is unlikely to have been damaged by the Ghibli episode and will likely carry on training itself on everything that has ever been created while improving its accuracy at attempting to materialize prompt-based hyperrealism or an affected style upon request.
Simon Manchipp, the brand strategy expert and founding partner of design agency SomeOne, calls OpenAI’s latest update “an ethical vacuum at the heart of generative AI.” He says: “This isn’t innovation, it’s imitation at scale, with Studio Ghibli’s reputation, built on decades of artistic mastery, reduced to a trendy AI gimmick.”
Manchipp poses the question: “How much longer will major brands risk the backlash that comes with AI-generated ‘art’? The very reason Studio Ghibli’s work is so beloved is because of its authenticity, human touch and intentional storytelling – qualities that AI simply cannot replicate.
“If OpenAI continues down this path unchecked, it risks more than just reputational damage. It risks creating a cultural landscape where audiences actively reject AI-generated content in favor of real artistry. And if brands have any foresight, they’ll recognize that partnering with real artists – not chasing the latest AI trend – is the only way to ensure their work remains valued and respected.”
Despite this unchecked development, ChatGPT is still failing to do what artists, animators and studios can, which is create a body of work with a human point of view, shown in all its subtlety and chaos.
Ultimately, this won’t halt its progress, though. This week, I looked at the legal landscape for creators and how copyright law is failing them in different regions around the world (except Japan, possibly). Opting out is the only option at the moment, but given that the large language models (LLM) have already studied most historical information, the damage is already done.
Studio Ghibli is now, like many others, at the mercy of LLMs. Jon Cockley is the co-founder of illustration agency Handsome Frank, which has the likes of Jean Jullien and Malika Favre on the books as well as offering animation services. His view is mixed, but he’s clear how “savage” the Ghibli trend is and how it is making illustrators “feel very exposed by it.”
On the point of the Ghibli ChatGPT style, Cockley reflects that “style is not just how you draw, the texture and the color palette, it’s also the subject matter, the idea and viewpoint, the angle and the use of light.”
As ChatGPT continues at such an aggressive pace without ethical considerations, Cockley, like Manchipp, feels that it will push some brands away from AI.
“What I’m seeing from premium and luxury brands – fashion, hotels and restaurants – is that they’re reacting to the flood of AI stuff by trying to commission work that looks and feels artist-led and clearly has the hand of the creator in it,” says Cockley, adding: “AI is mass market, it’s perceived to be cheap and that’s devaluing it.”
The Ghibli ChatGPT trend is “completely at odds with the trend on social media where artists are being required to step out from behind their work,” he says. This has led to a further investment in authenticity being made by clients. “They want work-in-progress videos, social content and for the artists to show their faces. The brand isn’t just investing in the artwork; it wants content to show it has worked with an artist, which AI obviously can’t do.”
This desire to be seen to be working with creators comes in tandem with a rejection of the aesthetic that AI often produces. The “AI look,” which he calls “a super-reality,” usually “isn’t for luxury clients” who are dismissing it as it feels faddy.
Cockley contributed to a recent UK government copyright review on AI and ownership but doesn’t hold out much hope on the legislative front, instead advocating for artist-owned models, which he holds a “guarded optimism” for.
Advertisement
I recently spoke with Tonia Samsonova of Exactly.AI, which gives creators ownership of their own model so they can use it for their own iterative purposes – on client work, for example – or make it public so they can be paid when a user makes something in their style.
Following the latest ChatGPT updates, Samsonova is unequivocal in her opinion that OpenAI’s approach is “reckless and unethical” and has forced many design agencies and publishers to adopt no AI policies.
She says: “I find it deeply troubling that a company that should be accelerating innovation is instead obstructing it. The release of OpenAI’s style transfer feature made it entirely predictable that users would start screen-shotting others’ work to generate derivative images. That’s not innovation, it’s theft. It enables the unauthorized replication of a creator’s life’s work, not just through foundational model training but also through mimicking styles from just a few images.
“Even more concerning is OpenAI’s claim that users have ownership over the content generated with their tools – a claim that completely ignores the rights of the original creators.”
She sees her platform as in direct opposition to OpenAI products. “We are actively cleaning up the mess created by OpenAI’s approach and showing the industry that ethical innovation is not only possible, it’s essential.”
Advertisement