With dueling summary judgment motions on illegal file sharing and fair use fully briefed, Judge Vince Chhabria is today the most important federal judge in all of the 41 lawsuits against AI companies in the United States. Given the timing, Judge Chhabria is poised to be the first federal judge to make a ruling on summary judgment regarding a fair use defense in the training of a generative AI model, here, involving Meta’s Llama, a large language model.*
Today’s hearing starts at 10 AM in San Francisco. As noted on Judge Chhabria’s page, you can listen by audio via Zoom:
REMOTE (ZOOM) ACCESS FOR MEMBERS OF THE PUBLIC. Audio of all in-person civil hearings and video of all Zoom civil hearings and case management conferences are available through the Zoom link below.
CASE PARTICIPANTS FOR MATTERS SCHEDULED TO BE HEARD BY ZOOM: Join using the Zoom link as an attendee, then you will be brought into the proceeding by court staff.
https://cand-uscourts.zoomgov.com/j/1612857657?pwd=WE5Gcm1zS293WU84V0tyd0c2Ulp1UT09
Webinar ID: 161 285 7657
Password: 547298
Judge Chhabria’s 12 Questions
Judge Chhabria’s list of 12 Questions for the parties to be prepared to answer today shows that the Judge has carefully scrutinized the parties’ respective arguments. And the Questions show that Judge Chhabria is troubled by some of the arguments by both sides.
Brief summary of Judge Chhabria’s questions, nearly all of which relate to fair use
- Factor 1: How to characterize’s Meta’s potential transformative use: (i) “because of what Llama is ultimately capable of producing” versus (ii) training an AI model in a way “different from a human reading the book so as to make it transformative”?
- Factor 1: If what the AI model is able to produce is relevant to whether it is a transformative purpose, is the fair use analysis affected because the model is able “to produce both comparable and non-comparable works” from the works on which it was trained?
- Factor 4: If the training involves every poem or magazine article ever written that enables the AI to write “innumerable” poems or articles, “hasn’t the use of the copyright-protected poems or articles to train the model made created a ‘substitute in a market the copyright holder reasonably expected?’”
- Summary judgment record: What does the summary judgment record say about the ability of Llama to do the things described above?
- Summary judgment record: If the record allows an inference that Llama can do the things described above, does the record say anything about how doing those things has affected or would affect the market for the copied works?
- Factors 2, 4 and summary judgment record: “Should works of fiction and non-fiction be thought of differently with respect to the fourth factor, in that it may be more likely for a language model to produce outputs that interfere with the market for original non-fiction works? Does the record say anything about this?”
- Failed transformative uses: “Are there cases where the secondary use was found to be transformative but the fair use doctrine was found not to apply (perhaps because of the fourth factor)?” Editor’s note: Yes, Warner Brothers v. RDR Books!
- Factor 1 and Relevance of Pirated Books: “There must be some difference, from a fair use standpoint, between downloading pirated works and using them to train AI versus lawfully acquiring the works and then using them to train AI. The plaintiffs’ argument that this is dispositive in their favor seems wrong, but Meta seems equally wrong to argue that it’s entirely irrelevant. Assuming the Court rejects both sides’ arguments on this point, how should the issue be treated?”
- Pirated Books follow-up question: “If companies are allowed to download pirated works to train their AI models, will that facilitate broader use of shadow libraries like LibGen? Does anything in the record speak to this question? How does the answer affect whether Meta’s downloading (and not just its alleged uploading) of copyrighted works is fair use?”
- Creativity of User of Llama and summary judgment record, and fair use: “In terms of promoting creativity through fair use, is it typically a person using Llama to do the creating? Or is it Llama itself that’s typically doing the creating? What does the record say about how Llama is commonly used in the real world? And does the distinction matter from a fair use standpoint?”
- Enhancing Llama’s creativity and effectiveness, and fair use: What does the record say about the extent to which using copyright-protected material enhances Llama’s creativity or effectiveness? Is this relevant to the fair use analysis?
- Close Case with Competing Interests, how to balance, ability to license: “Imagine a case where the evidence showed that: (1) allowing use of protected books to train an AI model would, to a degree, diminish the market for the copied works; and (2) disallowing use of protected books to train the AI model would, to a degree, diminish the effectiveness of the AI model’s ability to generate high-quality output. In that scenario, would the question of how easy or difficult it would be to obtain licenses to use the protected works for AI training be relevant to the fair use analysis?”
If the subject matter of these questions reflects how the Judge views the case, we might surmise that the Judge is more inclined to view Meta’s downloading of pirated books in connection with its AI training and fair use defense.
Notes
*Although Judge Bibas has already made a ruling rejecting ROSS Intelligence’s fair use defense for what he characterized as a non-generative AI model (because the model was trained to quote from judicial decisions instead of generating new content), that decision was rather short. Judge Bibas even allowed ROSS to petition for interlocutory appeal of his decision, given the substantial grounds for differing views on the issue.
Related Stories