Adapted by:
Jose García
WriterTech journalist. Head of new formats at Xataka and TikTok presenter. I specialize in consumer tech and video games.
Alba Mora
WriterAn established tech journalist, I entered the world of consumer tech by chance in 2018. In my writing and translating career, I've also covered a diverse range of topics, including entertainment, travel, science, and the economy.
It’s been a long time since an AI tool has generated so much excitement on social media. There’s usually quite a stir when a feature goes viral. However, the recent reaction to ChatGPT’s GPT-4o-based image generation is unprecedented. Generative AI has managed to do something it hasn’t achieved in a while: surprising everyday users. Unfortunately, it comes with one of AI biggest concerns: copyright infringement.
A viral trend. Social media platforms have recently been flooded with Studio Ghibli-style memes, images, and avatars created by ChatGPT. The results are truly impressive. However, you should remember that an AI tool like ChatGPT can generate an image of, say, a horse because it’s been trained on millions of images of horses.
Click the image to view the original post on X.
Context. If an AI model like ChatGPT-4o can create or transform an image in the style of a specific artist, it’s because it understands what that artist’s style entails. This means that ChatGPT-4o must have been trained on content related to, inspired by, or generated by Studio Ghibli.
While that content is beautiful, emotional, and relatable, it’s also not free or in the public domain. In other words, it’s copyrighted material. This issue has been a point of contention for ChatGPT and OpenAI since their beginnings.
No secret. ChatGPT was trained on a vast amount of data sourced from the Internet, including websites, books, social media posts, and academic articles. While much of this content may be freely accessible, it’s not necessarily free to use. An image that exists on the Internet is still hosted on a server that likely belongs to a company and is subject to copyright. Just because you can view and download an image for free doesn’t mean you have the right to print it for sale or use it to illustrate the cover of your next novel.
Click the image to view the original post on X.
“Living artists.” OpenAI says that it takes a “conservative approach” when it comes to generating images that use the work of other artists. The AI startup told Business Insider that it’s implemented “a refusal which triggers when a user attempts to generate an image in the style of a living artist,” like Studio Ghibli’s Hayao Miyazaki.
In response to the surge of images created in Miyazaki’s style, an OpenAI spokesperson also told the outlet that the company will block “generations in the style of individual living artists,” but will permit the generation of images in “broader studio styles.”
In other words, a user can’t create a Miyazaki-style image, but they can create a Studio Ghibli-style picture. This is somewhat ironic. In 2016, after viewing a demonstration of an AI-generated animation, Miyazaki said, “I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself.”
The style. No one can prevent someone from creating works in the style of Miyazaki or Studio Ghibli. Style itself isn’t protected by copyright. The real issue arises when protected works are used to train an AI to replicate that style. Similar to fan art, you’re free to create an illustration of Pikachu, print it, and display it in your room without any issues. However, you can’t sell that illustration.
An industry problem. The use of copyrighted content for commercial purposes has led to several complaints against OpenAI, with the most notable coming from The New York Times. Additionally, Getty Images filed a lawsuit against Stable Diffusion for using their images to train models. Anthropic faced legal action from a group of authors over the use of their books to train the Claude model. Moreover, Meta reportedly downloaded 81.7 TB of copyrighted books to use for training their models.
The cost of having access to AI seems to be the unchecked use of all content available on the Internet, despite AI companies claiming fair use. The prevailing notion among these companies is that if something is on the Internet, it’s free to use, which isn’t always true. All major AI companies have overlooked copyright laws, and until now, there have been no repercussions. AI copyright concerns are far from over.
Image | @heyBarsee
Log in to leave a comment