The complaint includes dozens of images that presumably demonstrate how Midjourney can evoke images with the intellectual property of the studies. An image depicts Yoda from Star Wars with a light saber in his hand, which according to him was made by inserting the prompt “Yoda with laser sword, IMAX”. Another shows that the typing of “The Boss Baby” as Prompt has led to an image of a child animated in a tuxedo that looks very like the protagonist of Universal The Child Chief franchise.
“This is an extremely significant development,” says Pi Chad Hummel’s lawyer, who sees the collection of images in the complaint as a convincing evidence that “the output is not sufficiently transformative”. Most of the artificial intelligence companies that face legal actions claimed to be protected by the doctrine of “fair use”, which allows the use of copyright protected works in certain circumstances; One of the main questions that the courts ask is whether the new work is “transformative” or adds a new meaning or message, when they determine the fair use.
Matthew Sag, professor of law and artificial intelligence at Emory University, believes that Midjourney will have difficulty making a case of fair use compared to previous artificial intelligence defendants.
“The reason why it is different is that Disney directly attacks the output of the model. It does not only use some examples collected by cherry to demonstrate that the model has been trained on its works,” he says. “It will be very difficult for a court or a jury to accept that it is transformative to take 1,000 images of Darth Vader and use them to produce even more images of Darth Vader.
The cause argues that Disney and Universal asked Midjourney to “adopt technological measures” to prevent its generators from images from producing materials in violation, but that the company has “ignored” their requests. In addition, it claims that Midjourney “clean” copies of Universal and Disney’s work during the training process, which “necessarily included the creation of multiple copies of materials”. Midjourney did not immediately respond to requests for comment.
“We are bullish on the promise of the Ai technology and optimists on how it can be used responsibly as a tool for further human creativity,” said Disney consultant to Horacio Gutierrez in a note. “But piracy is piracy and the fact that it is done by an artificial intelligence society does not make it less purple.”
Midjourney, like many other AI generative startups, has formed its tools by scraping the Internet to create large sets of images data, rather than looking for specific licenses. In a 2022 interview With Forbes, CEO David Holz has openly discussed the process. “It’s just a great scratch of the internet. We use the open data sets that are published and train through those,” he said. “There is really no way to get one hundred million images and know where they come from. It would be nice if the images had metadata incorporated in them of the copyright owner or something like that. But it is not one thing; there is no register.”