In the most high-profile case concerning the unlicensed and uncompensated use of billions of photos obtained from the internet to train AI systems, artists have achieved a little but significant victory in their fight against generative AI art generators. According to a federal court, the lawsuit is in the public interest, which refused to recognize that corporations can claim free speech rights.
In a judgment released last week, U.S. District Judge William Orrick rejected StabilityAI’s and Midjourney’s assertions that they had a First Amendment defense based on California legislation that permits the early dismissal of cases that try to limit free expression. Their position was that the lawsuit was an attack on their “speech” as it is a constitutionally protected activity to create art that expresses new ideas and concepts, such as the ones provided in text prompts to evoke hyper-realistic visuals.
The Stable Diffusion algorithm, used by Stability’s AI picture generator DreamStudio and supposedly by DeviantArt’s DreamUp and Midjourney, is the subject of a lawsuit that was filed last year in a federal court in California.
The court mostly agreed with the AI art creators’ motion to dismiss the case in October, although it did let several significant allegations forward. The court found that the accusations against DeviantArt and Midjourney regarding copyright infringement, right of publicity, unfair competition, and breach of contract are “defective in numerous respects, it decided not to pursue them. In order to recover its attorney fees and resolve the matter, DeviantArt moved for the decision on its motion to strike the claim to be rendered. This could have implications for future cases where AI companies claim First Amendment protections, even though a claim for the right of publicity was not reasserted when the suit was refiled. The artists then accused each other of wrongdoing. They emphasized that the corporations are trying to “strongarm and intimidate [them] into submission” by exploiting California’s anti-SLAPP statute.
A question that arose in the right of publicity case was whether or not AI art generators may advertise their wares using the names or styles of artists. The lawsuit stated that if the corporations are allowed to keep doing this, it will reduce demand for their original works.
By ruling that the “public interest exemption is met here,” Orrick took the side of the artists in the dispute over whether or not the corporations may dismiss the claim under the state’s anti-SLAPP legislation. He pointed out that at first, the claim was thrown out as the artists who filed the complaint—Sarah Andersen, Kelly McKernan, and Karla Ortiz—did not provide enough evidence to support the argument that the corporations had exploited their names in advertisements.
The court found that the plaintiffs would have made their claims “had plaintiffs been able to allege those facts,” according to the judgment. “That doesn’t discount the fact that their initial claims regarding the right of publicity were grounded in the fact that DreamUp was advertised or sold using their names. Such a claim would surely uphold California’s public policy that prohibits the appropriation of likenesses and names.” However,
In anticipation of discovery, the artists’ solicitors have retained the option to renew their right of publicity claim.
The majority of the lawsuit was rejected by the court in October, but a direct infringement claim against Stability AI might go on. This claim was based on claims that the business had unlawfully exploited copyrighted photos in developing its Stable Diffusion AI model. An essential part of its defense is the idea that training the chatbot doesn’t entail plagiarising works in their entirety but rather drawing on the works to define what things look like in terms of parameters such as lines, colors, shades, and other attributes related to subjects and concepts. The case’s outcome may depend on the result of the disputed issue.
After Midjourney’s most recent version in December stopped obfuscating what legal experts claimed was blatant copyright infringement, the app came under further scrutiny. The AI art generator might be instructed by users to produce virtually identical reproductions of scenes from famous films. The purple-skinned villain from “Thanos Infinity War” emerges in a shot that seems to have been pulled directly from the movie or promotional materials when Midjourney was presented with that name. The chatbot can also produce visually stunning characters from a variety of films, including Shrek (DreamWorks), Ratatouille (Pixar), and The Lego Movie (Warner Bros.), apparently mimicking the animation techniques of these films.
Leave a Reply