The debate behind Open AI’s viral images

In a matter of hours, Premium of Chatgpt User They turned the Internet into a gallery of illustrations inspired by Studio Ghibli, from pet portraits to reproductions of historical moments -like the attacks of the Twin Towers or the painting of the Spanish painter Francisco de Goya of ‘The executions of May Tres de Mayo’-, passing through iconic memes, such as’ The distracted boyfriend’.

However, the trend also highlighted ethical concerns about artificial intelligence tools trained with creative works protected by copyright and what that means for the future support of human artists.

Just looking for the term Studio Ghibli on social networks Instagram or X will immediately appear on the screen an endless iconic images, such as the Casual appearance of Turkish shooter Yusuf Dikec or the famous meme “Disaster Girl” From a four -year -old girl turning to the camera with a slight smile while a fire in a house burns in the background.

The creator of Chatgpt, Openai, who faces demands for copyright on his flagbot flagbot, greatly encouraged the “ghiblification” experiments and its executive director, Sam Altmanhe changed his profile on the social media platform X to a Ghibli -style portrait.

The recent incorporation of advanced capabilities in the generation of images in the GPT-4 model caused great enthusiasm in users. However, this enthusiasm caused different problems to Openai and led them to have to implement some temporal limits In image generation applications.

Sam Altman, CEO of the company, said through the social network X: “It is very fun to see people enjoy the images in Chatgpt, But our GPU are melting“.

The CEO warning opted in a technical document published last Tuesday, in full viral fury. His company said the new tool would take a “conservative approach” in the way in which the aesthetics of individual artists mimic. “We have added a refusal that is activated when a user tries to generate an image in the style of a living artist,” they said.

Specifically, every time Chatgpt is requested to generate an image of this type, now shows a warning: “OpenAi imposes urgent limits in the generation of images due to a high level of demand and possible related problems“.

However, the company added in a statement that “allows broader study stylesthat people use to generate and share some original creations of truly lovely and inspired fans. ”

For his part, Studio Ghibli in Japan declined to comment on Friday.

As users published this type of images on social networks, Miyazaki’s previous comments about AI animation also began to resurface.

When the Japanese artist, known for his hand drawing approach and his imaginative narratives, a demonstration of AI was presented in 2016, said that I was “totally disgusted” by the exhibitionaccording to documentary images of the interaction.

The person who demonstrated animation, who had a body twisting and crawling down the head, explained that AI could “present grotesque movements that humans cannot imagine.” It could be used for zombie movements, said the person.

That led Miyazaki to tell a story: “Every morning, not in recent days, I see my friend who has a disability,” Miyazaki said. “It is so difficult for him to simply make a hand greeting; his arm with rigid muscle cannot reach my hand. Now, thinking about him, I can’t see these things and find them interesting. Who creates these things has no idea what pain is”.

He said that “I would never want to incorporate this technology in my work at all.” And it was for more: “I strongly feel this It is an insult to life itself”.

Josh Weigensberg, a partner of the law firm Pryor Cashman, said that a question that raises the art of the Ghibli style is whether the model was trained in the work of Miyazaki or Studio Ghibli. That in turn “raises the question of, ‘well, Do you have a license or permit to do that training or not?‘”, said.

In turn, the lawyer added that if the rights of a work were ceded for training, it could make sense that a company allowed this type of use. But if this type of use is happening without consent and compensation, he said, could be “problematic”.

Weigensberg also said that there is a general principle “in bird view” that the “style” is not protected by copyright. But sometimes, he acknowledged, what people are really thinking when they say “style” could be “more specific, discernible, discreet elements of a work of art,” he said.

“In the case of ‘The incredible vagabond castle’ (” Hauru no Ugoku Shiro “) or ‘Chihiro’s trip’, you could freeze a frame in any of those films and point out specific things, and then look at the result of the generative AI and see identical or substantially similar elements,” he said. “Simply stop in, ‘Oh! Well, The style is not protected under the copyright law‘. That is not necessarily the end of the investigation. ”

The artist Karla Ortiz, who grew watching Miyazaki’s films and demanded other generators of images of AI due to copyright infractions in a case that is still pending, described it as “another clear example of how companies like OpenAi simply simply They do not care about the work of artists and the sustenance of the artists. ”

“That is using the Ghibli brand, your name, your work, your reputation, to promote (OpenAi) products,” Ortiz said. And he added: “It’s an insult. It is exploitation”.

Ortiz was further enraged when the administration of President Donald Trump joined the memes trend on Thursday, using the official X of the White House to publish a Ghibli -style image of a woman from the Dominican Republic crying after being recently arrested by United States immigration agents. The White House and Openai did not give details about how the image was made.

“Seeing something as bright, as wonderful as Miyazaki’s work to be shattered to generate something so vile,” Ortiz wrote on social networks, adding that he expected Studio Ghibli will sue “with everything” to OpenAI For this.