In Sarah Andersen v. Stability AI, Judge Orrick issued his order granting in part the defendants’ motion to dismiss the DMCA CMI claims, contract claims, and breach of implied covenant of good faith and fair dealing (against DeviantArt) with prejudice and unjust enrichment claim with leave to amend, but denying the motion on the key claim of copyright infringement, as well as the Lanham Act claims against Midjourney.
All told, this is a big win for Sarah Andersen and the other artist plaintiffs. A very big win. This was the first class action copyright lawsuit filed against AI companies. Judge Orrick had dismissed the original complaint in terms of the allegations of infringing output from the AI models. The plaintiffs amended their complaint to add further allegations in the first amended complaint sufficient to allow the copyright claims to proceed, in Judge Orrick’s view, both on direct infringement in terms of the alleged “compressed copies” of training images and a theory of active inducement. (The claims based on the training copies, of course, remain too — the defendants did not include them in their motion to dismiss (because those allegations no doubt state a claim)).
Although Judge Orrick repeatedly mentioned in his order that the plaintiffs’ claims can be tested at the summary judgment stage, the Judge found the plaintiffs’ allegations of infringement plausible to survive a motion to dismiss, meaning if the allegations are assumed to be true for the purpose of reviewing the motion, they did properly state a basis for copyright liability and liability as to Midjourney under the Lanham Act. That includes not only the plaintiffs’ “compressed copies” theory (that the AI models have compressed copies of training images stored in them, which enables them to recreate the training images in whole or in part during use of the model), but also to the plaintiffs’ theory of active inducement of infringement by others. Wow.
Below are highlights of Judge Orrick’s discussion of the federal claims for copyright and Lanham Act violations for false endorsement and vicarious trade dress infringement.
Judge Orrick’s analysis of Stability AI’s motion to dismiss copyright claim
However, plaintiffs point to one statement by Stability’s CEO that Stability took 100,000 gigabytes of images and compressed it to a two-gigabyte file that can “recreate” any of those images. Stability responds that the “isolated” use of the word “create” by its CEO cannot demonstrate that it intended to foster infringement. Stability argues that “clear allegations of active steps to encourage direct infringement” are especially important here, where plaintiffs do not dispute that Stable Diffusion is capable of substantial noninfringing uses, like creating art from inputs that do not reference particular artists or invoke particular artists’ styles or have any substantial similarity to plaintiffs’ works.
The theory of this case is not similar to – for example – a case asserting contributory infringement based on the sale of VCRs where, after discovery, plaintiff had no evidence of defendant’s intent to induce infringement. The Supreme Court explained that, in those circumstances, intent could not be “based on presuming or imputing intent to cause infringement solely from the design or distribution of a product capable of substantial lawful use, which the distributor knows is in fact used for infringement.” See Grokster, Ltd., 545 U.S. at 933 (discussing holding of Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417 (1984)). Instead, this is a case where plaintiffs allege that Stable Diffusion is built to a significant extent on copyrighted works and that the way the product operates necessarily invokes copies or protected elements of those works. The plausible inferences at this juncture are that Stable Diffusion by operation by end users creates copyright infringement and was created to facilitate that infringement by design. In addition to the comment of Stability’s CEO, plaintiffs reference articles by academics and others that training images can sometimes be reproduced as outputs from the AI products. FAC ¶¶ 90, 130-139.
Whether true and whether the result of a glitch (as Stability contends) or by design (plaintiffs’ contention) will be tested at a later date. The allegations of induced infringement are sufficient.
Judge Orrick’s analysis of Runway’s motion to dismiss copyright claim
Plaintiffs addressed the deficiencies in their prior complaint, alleging additional facts in the FAC concerning how the training images remain in and are used by Stable Diffusion. See, e.g., FAC ¶¶ 71, 83, 88-90, 150.15. Plaintiffs rely, with respect to Runway as with Stability, on comments from Stability’s CEO regarding the contents of the model and their ability to reproduce works, as well as academic papers indicating that the Stable Diffusion models are capable of producing very similar if not identical works to at least some training images. FAC ¶¶ 122-150. And plaintiffs also rely on use of their names as prompts in the Runway products to create outputs mimicking aspects of plaintiffs’ protected works as evidence that their protected works are being copied or distributed in Runway’s product. Id. ¶¶ 163-169.
Runway disputes the accuracy of those assertions – including the full meaning and import of the academic articles relied on by plaintiffs – and argues that the prompt-examples do not support assertions of direct copyright infringement absent express identification of outputs that are “substantially similar” to plaintiffs’ copyrighted works. But the allegations at this juncture are sufficient to allow the direct infringement claims to proceed. Whether evidence can support each of the theories and whether plaintiffs will need to choose between theories (e.g., between direct infringement based on selling a product containing effective copies of copyrighted works or violating plaintiffs’ rights to restrict distribution of their works) will be addressed at summary judgment.
Runway’s motion to dismiss the direct infringement claims is DENIED.
Judge Orrick’s analysis of Midjourney’s motion to dismiss copyright claim
[P]laintiffs have added to their FAC more detailed allegations regarding the training and use of the LAOIN datasets by defendants generally and Midjourney specifically. Plaintiffs have plausible allegations showing why they believe their works were included in the LAION datasets. And plaintiffs plausibly allege that the Midjourney product produces images – when their own names are used as prompts – that are similar to plaintiffs’ artistic works. See FAC Exs. F&G.
Midjourney nonetheless argues that these examples are insufficient because some of the identified works are not registered and the resulting outputs could just as likely be the result of training on unregistered works or utilizing only unprotected elements from plaintiffs’ works. But plaintiffs’ reliance on those exhibits are not to establish copyright infringement as a matter of law. Instead, they are relied on to support the plausibility of plaintiffs’ copyright theories (that all or most of the works in the LAION datasets were used by Midjourney and the other defendants to train their AI products, and that plaintiffs’ works or their protected elements that are contained in the AI products as the works or protected elements can be recreated by using the AI products). The FAC allegations and the exhibits help plaintiffs cross the plausibility threshold. Whether plaintiffs will be able to prove their claims is a different matter and those claims will be tested on an evidentiary basis at summary judgment.
Midjourney’s motion to dismiss the Copyright Act claims is DENIED.
Judge Orrick’s analysis of Midjourney’s motion to dismiss Lanham Act claims of false endorsement and vicarious trade dress
In place of the right of publicity claims asserted against Midjourney that were dismissed in my October 2023 Order, five plaintiffs (Anderson, Brom, Kaye, Ortiz and Rutkowski) now assert Lanham Act claims based on theories of false endorsement and trade dress. These plaintiffs allege that their names appeared on the list of 4700 artists posted by Midjourney’s CEO on Discord, the platform where Midjourney’s AI product operates. Midjourney’s CEO promoted the list as describing the various styles of artistic works its AI product could produce. FAC ¶¶ 261-262, Plaintiffs also allege that Midjourney has itself published user-created images thatan evidentiary basis at summary judgment. incorporate the plaintiff artists’ names in Midjourney’s “showcase” site. Id. ¶ 325(b) & Ex. K.
As discussed above, judicial notice of other comments in the thread is not appropriate todispute the facts plaintiffs otherwise plausibly assert, especially as the plaintiffs dispute the accuracy and inferences to be drawn from the totality of the messages in that thread. And even if the Names List itself was insufficient to support an inference of false endorsement, plaintiffs also allege that their names were used in connection with works included in Midjourney’s “showcase.” FAC ¶ 325b. Whether or not a reasonably prudent consumer would be confused or misled by the Names List and showcase to conclude that the included artists were endorsing the Midjourney product can be tested at summary judgment. See Adobe Sys. Inc. v. A & S Elecs., Inc., 153 F. Supp. 3d 1136, 1142 (N.D. Cal. 2015).
Midjourney also argues that to get past the First Amendment protection provided to expressive works, plaintiffs must, but have not, alleged that Midjourney’s use of plaintiffs’ names to invoke their styles has “‘no artistic relevance to the underlying work whatsoever, or, if it has some artistic relevance, unless the [use of trademark or other identifying material] explicitly misleads as to the source or the content of the work.’” Brown v. Elec. Arts, Inc., 724 F.3d 1235, 1239 (9th Cir. 2013) (quoting Rogers v. Grimaldi, 875 F.2d 994 (2d Cir.1989)). However, the plaintiffs have plausibly alleged that the use of their names by Midjourney in the List and showcase misleads consumers regarding source and endorsement. Open questions also remain whether Midjourney promoting its product for commercial gain for use by others to create artistic images is itself expressive use that creates “artistic relevance” to plaintiffs’ underlying works. Discovery may show that it is or that is it not. Unlike in Brown v. Elec. Arts, Inc., where plaintiff alleged only that his likeness was used in a video game, we do not yet have the sort of record, or sort of alleged use, that could support dismissal of the claim at the motion to dismiss stage.
The case now will proceed to discovery. In addition to Sarah Andersen, the other artist plaintiffs are
Kelly McKernan, Karla Ortiz, Hawke Southworth, Grzegorz Rutkowski, Gregory Manchess, Gerald Brom, Jingna Zhang, Julia Kaye, and Adam Ellis. They are represented by the Joseph Saveri law firm, which is involved in a number of the copyright lawsuits against AI companies.
Download the entire order below: