The extremely popular chatbot tool of Openai, Chatgpt, attracted a huge wave this week when indigenous numbers began to use it to generate images evoking the style of the legend of animation Hayao Miyazaki at the Ghibli studio.
The trend quickly took its own life. Even OpenAi CEO Sam Altman went to X-Formly-Twitter to joke about being transformed into “Ghibli Twink style.
Openai has since started to repress the trend, implementing a confusing range of changes which now often refuse requests for generation of images in the style of Ghibli.
The viral trend and chaotic response of Openai have rekindled a stormy debate surrounding copyright and the use of a generative AI directly undergoing the work of human artists and publishers – including the extremely talented animators of Ghibli.
And according to the former lawyer general of the television channel Showtime and the expert in IA Rob Rosenberg, Ghibli could in fact be able to pursue legal proceedings against Openai on the situation.
Although it is a “complex question” of knowing whether the Japanese animation studio would have sufficient reasons, the company “could have the capacity to claim that Optai has violated the Lanham law which provides the basis of complaints related to false advertising, the violation of brands and unfair competition,” said Rosenberg, now the founder of Legal Strategies, said Futurism.
The Lanham Act, which was promulgated in 1946, is the main federal law governing brands law in the United States and establishes a national brand registration system which allows owners to pursue prosecution against counterfeiting.
“Ghibli could say that by converting user photos into” Ghibli-style “, OpenAi exchanges the goodwill of Ghibli’s brands, using Ghibli’s identifiable style and leading to a probability of confusion among consumers that this function is approved or licensed by Studio Ghibli,” said Rosenberg.
Openai could also be prosecuted to “train his large language models on the works protected by the copyright of Studio Ghibli”, he added.
OPENAI is already faced with several proceedings on its approach to browse Internet for data training for its models, including many documents protected by copyright.
This week, a federal judge rejected OpenAi’s request to take action in copyright filed by the New York Times. The newspaper said at the end of 2023 that the company managed by Altman had violated its copyright using its equipment without authorization to train its AI models.
Mark Zuckerberg’s meta has also been accused of illegally downloading pirated books to form its AI models.
But so far, there is no legal precedent to determine if Openai has violated the law on copyright. The company argued that “the training of their models is fair use under the law on copyright”, explained Rosenberg “, and therefore an exception to violation”.
The “doctrine of fair use” has been fundamental to the current debate surrounding if IA societies as Openai violate the copyright law. This is a legal framework allowing limited access to documents protected by copyright without prior authorization for quotes and other non -infiltrated uses.
The question of whether Ghibli will take legal measures is someone’s assumption. Friday, the company told a Japanese media NHK That a viral legal document, which claims to be a letter to stop and abstain sent by the studio, was false.
“As most of the courts have not yet ruled on the strength of similar complaints on copyright, it is not clear whether they are successful or not,” said Rosenberg Futurism. “If Openai marketed its platform as a place where consumers could go to transform their photos into an” Ghibli-style anime “, this fact would weigh more strongly in favor of a court according to which the conduct of Openai violated the Lanham law.”
Unless the courts were to “governing unanimously” in favor of OpenAi, to determine that the training of AI models on the material protected by copyright falls in fair use, we could lead “to a system where IA developers will have to compensate for copyright holders and credit them when their content is used in production production”, said Rosenberg.
Aside from the questions concerning the possible violation of copyright, Openai could steal Ghibli in other ways to monetize.
“If Studio Ghibli has always wanted to launch its own tool allowing fans to transform photos into its signing style, Openai’s update has essentially taken this commercial opportunity,” said Rosenberg. “Worse, this technology could lead people to generate whole animated projects in the Ghibli style and to present them falsely as authentic.”
“It is similar to ask the AI to rewrite a news in the style of JK Rowling, then to publish it online like a work of Rowling” lost “, he added,” blurring the border between homage and false declaration. “
More on the Copyright: The cool site shows exactly which books downloaded from Zuckerberg Minions to form Meta’s Ai
