AI generated art leaves us with fuzzy boundaries

Artists, animators, and fans have rallied in defense of Studio Ghibli's creative rights.

By Deepak Narayanan 

When OpenAI dropped the new version of its image generator last month, the internet promptly split into two camps: Camp 1 articulately explaining why this was nothing short of art theft, and Camp 2 spamming our feeds with Studio Ghibli-style self-portraits.

The Japanese animation studio's distinctive style — painstakingly crafted over decades by an army of artists led by Hayao Miyazaki — had been absorbed, processed, and regurgitated by machine learning systems without permission, compensation, or acknowledgment. Artists, animators, and fans rallied in defense of Studio Ghibli's creative rights, condemning what they saw as straightforward theft.

Shades of grey

But as we navigate this new world where AI increasingly intersects with human creativity, professional expertise, cultural traditions, and our varying ideas of progress and innovation, the answers aren’t going to be as black-and-white as we’d like.

While the Ghibli situation struck many as an obvious transgression, we're rapidly entering territory filled with expanding shades of grey — areas where competing values and legitimate interests will create genuine ethical dilemmas.

This column explores where AI usage raises complex ethical questions without obvious answers. The goal isn't to prescribe solutions but to invite deeper thinking about how we balance innovation with existing rights, and social benefits with individual protections.

“Inspired” to “copied from”

Let’s go back to the Ghibli controversy for a moment, to try and see at what precise point AI crossed the line from “inspired by” to “copied from”.

* A model trains on thousands of artists’ works and starts generating dreamy, whimsical imagery — reminiscent of Ghibli, but nothing directly lifted.

* It begins to consistently echo specific stylistic quirks: the cloud shapes, colour palettes, the way wind moves through grass — familiar, but arguably homage.

* Now it’s churning out portraits with the same big, soft eyes and flushed cheeks that feel lifted straight out of Spirited Away — but they’re of your friends. It’s fun.

* You find a near-identical frame from a real Ghibli film — same composition, same lighting, just AI-warped and rebranded.

It’s easy to classify the last one as theft, but you can see there’s a sliding scale involved.

For years, we've had apps that turned your selfies into anime characters. They've also been around as fun, novelty filters on apps like Snapchat and Instagram, or on big design apps like Canva and Fotor. What made all of that unworthy of the outrage we’ve seen recently?

To put it more bluntly - is it ok to use AI to mimic styles of lesser-known artists just because they aren’t as famous as Miyazaki?

Let’s take another example, this time hypothetical:

# A climate research nonprofit is facing a problem. Despite producing rigorous, data-backed reports, their social media posts get zero engagement. They’re unable to rally support for their cause, and keep hitting a wall when they need to raise money. So they try something new.

# They partner with a team of AI researchers to build a content-generation system designed not just to inform, but to influence. They train a model on thousands of public social media posts from popular influencers — not climate activists, but lifestyle creators. People in fashion, food, fitness, design — the kinds of voices that shape trends and habits across diverse, global audiences. The model learns how these influencers speak: their phrasing, pacing, emoji choices, even how they structure a caption to drive clicks or spark comments.

# Their research demonstrates that these AI-mimicked styles significantly increase public engagement with climate science among diverse demographic groups compared to generic messaging.

This sounds like a reasonable use of AI, right? Climate change represents an existential threat requiring unprecedented mobilization of resources and attention. If certain communication styles demonstrably improve public understanding and motivation to act, shouldn't we use them?

The nonprofit isn't profiting commercially, and they're addressing an urgent global crisis where effective communication could literally save lives. The AI isn't copying specific content but rather adopting structural elements and tonal characteristics that make information more accessible to different audiences.

Back to where we started: Where do we draw the line?

Having said that, creative styles represent years — sometimes decades — of a writer's development, experimentation, and personal expression. An argument can be made that their voice is unique, and is inseparable from their identity as creators.

Also, who decides which causes qualify? Today's climate crisis could be tomorrow's political campaign. These questions will either grow more urgent and complicated — or quietly fade from our attention entirely.

Technology will continue to move faster than legislation (and news cycles even faster), leaving us to navigate this terrain with imperfect tools and evolving norms. In fact, we could argue that if AI eventually handles most tasks, routine and otherwise, our distinctly human contribution may become our capacity for moral reasoning and ethical judgment.

Which brings us to the original question: when it comes to AI generated art, where do we draw the line?

(Deepak Narayanan spent two decades as a journalist before moving into the world of tech. He is the co-founder of chopchop.video.)

Views are personal and do not represent the stand of this publication.