Millions of people watch ChatGPT as it invents images that closely resemble Studio Ghibli

Millions of people watch ChatGPT as it invents images that closely resemble Studio Ghibli

Millions of people are currently excited about ChatGPT and its ability to create images in the Ghibli style.

Anyone currently looking at X.com or other social media will find that many images resembling the style of Studio Ghibli can be found. This is because ChatGPT-4o can generate content in the style of a specific author.

However, the problem is that ChatGPT must have purchased this advantage at the cost of serious copyright violations: it seems that ChatGPT-4o has been trained with content that is associated with, based on, or created by the studio founded by Hayao Miyazaki. And AI like ChatGPT has long been suspected of copying material.

ChatGPT can create images in the Ghibli style, but it involves protected material

What is the problem? Material or films such as those from Studio Ghibli are protected by copyright. The rights to these films are not in the public domain and therefore cannot be used freely by anyone. This is a problem that has plagued ChatGPT and OpenAI since their inception.

ChatGPT has been trained with an enormous amount of data from the internet – from websites, books, social media posts to academic articles. Content that is freely accessible but not free. Not every image you find on the internet can you use for free unless the author/photographer explicitly releases their work under a Creative Commons license.

However, it is also important to note that no one can stop others from creating works in the style of a person or a studio. Because the style itself is not protected. This is similar to the discussion about fan art: you can create your own image of a Pokémon and hang it on your wall, and no one will mind. However, as soon as you start trying to sell these artworks, it gets legally complicated.

AI companies generally rely on the “Fair Use” doctrine for their data, which is particularly applicable in the US. Just recently, a US court ruled that training AI with copyrighted content does not automatically fall under the Fair Use doctrine (via advomare.de). However, it is important to note: the ruling only concerns non-generative AI and is therefore less relevant for AI models like ChatGPT.

What is OpenAI doing about it? OpenAI claims to have implemented a ban to prevent the creation of images or artworks that aim to replicate the style of a living artist. Deceased artists, on the other hand, are acceptable. It is also okay to copy “studio styles,” which presumably includes Ghibli Studios (via xataka.com).

Lawsuits Against OpenAI and Demands for Protected Material

How do artists and authors view this? OpenAI and other AI companies have already been sued for serious copyright violations. This includes artists and image databases that are opposing OpenAI using their content for free, even though it is not free. Among others, 17 prominent authors, including George R. R. Martin, have joined a lawsuit against ChatGPT.

The authors’ accusation: the AI learns from texts that are illegally downloaded from the internet. In a statement at that time, it was said that “generative AI threatens to destroy the profession of authors.”

By the way, OpenAI is currently using this as an opportunity to demand access to protected materials. They believe that if they do not gain access, China will win the AI race: OpenAI says they urgently need access to legally protected material – because of China

Deine Meinung? Diskutiere mit uns!
2
I like it!
This is an AI-powered translation. Some inaccuracies might exist.
Lost Password

Please enter your username or email address. You will receive a link to create a new password via email.