Kadrey v. Meta is one of the most important copyright lawsuits involving AI. Judge Chhabria will hold oral argument on the parties’ motions for partial summary judgment on May 1, 10:00 AM in San Francisco federal court.
We prepared a new infographic to summarize the main points each side is advancing in their motion for partial summary judgment. Kadrey and Meta have vastly different theories of how Judge Chhabria should view the (alleged) conduct of Meta. We will take a closer look at these legal arguments in a subsequent post.
- Kadrey’s main theory: The summary judgment motion focuses on Meta’s putatively illegal file sharing. Meta’s file sharing and downloading of millions of copies of “pirated” books from illegal books datasets online offered by notorious sources investigated by the FBI should be analyzed as Meta’s separate use of plaintiffs’ books (which were in some of the “pirated” books datasets Meta obtained). As a separate use by Meta, these illegal copies obtained from unauthorized file sharing, including sharing or “leeching” copies of the books with third parties while Meta torrented the files, and downloading constitute infringement. Whether Meta made a fair use afterwards to train their AI model is another separate use by Meta that should be decided at trial because factual disputes related to the four fair use factors remain.
- Meta’s main theory: The summary judgment motion focuses on Meta’s putatively transformative purpose in AI training, which was the reason why it downloaded and torrented the books datasets. Meta’s downloading of books datasets from online sources was for the transformative purpose of developing a new technology, its AI model Llama, which is a quintessential fair use purpose. (Meta argues downloading should not be considered a separate use as Kadrey argues, it was for Meta’s purpose in developing an AI model.) All four factors favor fair use. Meta disputes that Kadrey’s motion for summary judgment applies to “leeching,” which Meta characterizes as a distribution claim, not reproduction claim–the only thing at issue in Kadrey’s motion, according to Meta. Moreover, Meta disputes there is any evidence Meta ever made an actual distribution of the plaintiffs’ works during “leeching.”
- Kadrey alleges Meta “destroyed” logs related to Meta’s file sharing but Meta says it used direct download (not torrenting with leeching) for books dataset with plaintiffs’ books. A secondary issue involves whether Meta overwrote any electronic logs that would have indicated which books Meta may have shared copies of with third parties during the so-called “leeching” process of torrenting. Kadrey asserts Meta’s overwriting of electronic logs is the reason why it cannot identify specific titles of books Meta shared. Meta denies that allegation and insists it direct downloaded the dataset(s) containing the plaintiffs’ books (in which case there was no “leeching” for those books).