Because it can
AI generated image, any similarity to actual persons, living or dead, is purely coincidental
In my previous article “If a tree falls in a forest” I highlighted two new Large Language Models (LLMs) that overshadowed other LLM releases due to their widespread adoption and people sharing their results online. One of these models became particularly famous for its ability to mimic renowned artistic styles, even coining new terms and trends – notably, Ghiblification. You might have guessed it: I’m referring to 4o Image Generation by OpenAI.
Ghiblification involves using LLM to transform images into a style reminiscent of Studio Ghibli’s distinctive anime art. This neologism first appeared in online discussions last month, March 2025, with the earliest recorded use being an X post by user @suntzoogway on 26 March 2025. In this post, @suntzoogway advocates for preserving original media, such as music tapes and film DVDs, to prevent Big Tech from monopolizing historical data.
Reflecting on various historical methods of storing knowledge, as discussed in my other article “Think, Write, Innovate”, I struggle to agree with this perspective. Do we really want to preserve clay tablets or DVDs as primary sources of knowledge for future generations? Or is the real problem here different? Evidence suggests that Big Tech not only transfers prior knowledge but also inserts their own biases, intentionally omitting certain historical facts to favor their preferred interpretations.
As a result, there is a new spin to the older notion that if you are not on the Internet then you do not exist. First, if LLM was not trained on your data, you risk fading into obscurity. Second, even if you are online and LLM was trained on your data, will LLM portray you accurately?
Returning to Ghiblification, the ethical question is whether anyone should create and use LLMs simply because they can. Research indicates that OpenAI did not sign an agreement with Studio Ghibli. Legal questions arise about whether OpenAI trained its LLM on Studio Ghibli's copyrighted works, including derivative works available online, thereby intentionally infringing on their copyright by releasing the LLM. In my opinion, the fair use doctrine, in its classical interpretation, hardly applies here, as OpenAI's actions were clearly for commercial purposes rather than limited research. To be fair to OpenAI, you can insert a name of one of other dozens of companies that provide similar Ghiblification capabilities and ask yourself the same questions.
Without diving into complex AI-related copyright debates and acknowledging that copyright law is not a law of nature and will be amended to new social norms, the bigger question to me personally is what those new social norms would be. Is it acceptable to create and use new LLM simply because it can be done? How do you feel about Google deciding to remove limitations on using AI for military purposes on 4 February 2025?
What are your thoughts on Ghiblification as a cultural phenomenon and the broader implications of the "because it can" attitude? Should we follow the trend and Ghiblify every image, or resist the temptation and support Studio Ghibli by purchasing their products? Please share your thoughts in the comments.
This article was written for fun, please do not judge. Instead, please share your comments in a constructive and respectful manner. The author and AI remain innocent until proven guilty.
