, , ,

Bombshell allegations: Kadrey plaintiffs assert “Meta’s in-house counsel engaged in and advised Meta on Its Illegal Activity” and invoke crime-fraud exception for discovery. Also claim Mark Zuckerberg should be deposed again.

Discovery has closed in Kadrey v. Meta. But the plaintiffs are now seeking not only further discovery, including second depositions of key Meta witnesses, including Mark Zuckerberg, in light of what plaintiffs allege is new information revealed late in discovery, but also leave to file a Third Amended Complaint to assert (again) DMCA CMI claims and California Comprehensive Computer Data and Access Fraud Act (CDAFA) claims.

Magistrate Judge Hixson will be deciding these various discovery disputes.

A. Crime-Fraud Exception

Even more stunning, in a flurry of discovery dispute letters filed on December 20, the plaintiffs are now asserting “Meta’s In-House Counsel engaged in and advised Meta on its illegal activity”:

Meta (downloaded and distributed) copyrighted works in pirated datasets
(violating CDAFA), stripped works of CMI to conceal their copyrighted status (violating the DMCA), and used them to train Llama (violating the Copyright Act). Meta engaged its in-house counsel to approve, conceal, and justify this illegal scheme. An NYT article produced earlier in discovery, entitled “How Tech Giants Cut Corners to Harvest Data for A.I.,” describes significant in-house counsel involvement in Meta’s copyright infringement scheme. Dkt. 262-8. As the Times explained, “some of the company’s business development leaders, engineers and lawyers met nearly daily . . . . [and] talked about how they had summarized books, essays and other works from the internet without permission and discussed sucking up more, even if that meant facing lawsuits.”
Id. at 8. Recordings of internal Meta meetings made clear that in-house counsel was not only aware that Meta was using copyrighted data, but that they were involved in planning how to do so.”

Plaintiffs’ position

But the plaintiffs relegate to a footnote Meta’s potential fair use defense:

1 Meta’s fair use defense is not relevant to the crime-fraud inquiry. Its argument that this was fair use after the fact cannot negate a prima facie showing of crime-fraud. If the mere fact that Meta raised a defense to Plaintiffs’ claim was sufficient to defeat the crime-fraud exception, then the exception would be overwhelmed. Any abuse of the attorney-client relationship could be concealed simply by raising a defense.

Plaintiffs’ position

Invoking the crime-fraud exception, plaintiffs seek documents from Meta that it asserted were protected by attorney-client privilege.

Meta responds that the assertion of crime-fraud exception is baseless:

Nowhere have Plaintiffs alleged, nor could they, that Meta did not reasonably believe that its downloading and use of certain datasets for AI training is fair use–an issue that is being tested in multiple lawsuits across the U.S. Critically, to date no court or other authority has issued an opinion regarding the legality of the conduct that Plaintiffs brand as a crime or fraud: gathering data from the internet for training AI models, including through ; removing copyright information (along with many other categories of repetitive information) from data prior to AI training to improve model performance; and/or training AI models on copyrighted data. To the contrary, this Court and the U.S. Copyright Office3 have recognized that whether fair use applies to the conduct alleged in this case is a novel question. Plaintiffs certainly have not established such use was not fair. Indeed, even the New York Times article that Plaintiffs cite as supposed support
for their crime-fraud allegation describes Meta’s genuine belief that its use of copyrighted works is fair. Accordingly, Meta employees’ efforts to work with counsel to navigate these issues can in no way be viewed as furthering any unlawful scheme.

3
“Perlmutter told lawmakers fair use is ‘the most controversial of all the areas that we’re
addressing” in a forthcoming report. Ivan Moreno, Copyright Chief: Fair Use A Tough Issue In
Upcoming AI Report, LAW360 (Nov. 13, 2024), https://www.law360.com/articles/2258152/copy
right-chief-fair-use-a-tough-issue-in-upcoming-ai-report.

Meta’s position

B. Mark Zuckerberg deposition again

Plaintiffs also want to depose a second time some of the Meta witnesses, including Mark Zuckerberg, due to what plaintiffs assert was a “massive, eleventh hour document dump” from Meta that provided putatively incriminating information that plaintiffs would like to depose the witnesses about. Plaintiffs assert that Meta had collected some of these documents back in June and September 2024, but didn’t produce them until just recently.

From the plaintiffs’ statement:

Plaintiffs’ position

These documents do not merely add color to the deposition testimony. Instead, they reflect repeated instances where answers in deposition were directly contradicted by documents. That is why Plaintiffs request further deposition time. And that includes Mr. Zuckerberg, who should be questioned, as Meta’s CEO, about

As it stands, because no other document (and, conspicuously, not a single witness) speaks to it is especially important that he be questioned about that document. The document directly implicates him, but was not in his custodial files. Plaintiffs did review the new Zuckerberg custodial documents in time for his deposition, but this document was not one of them—it was one of the 2,300 other new documents.

Plaintiffs’ position

Meta says it didn’t hold back anything, but instead, plaintiffs were responsible for the delay in discovery requests.


Plaintiffs incorrectly argue that Meta held back certain documents that should have been produced earlier, and thus would have been available for the six depositions they now seek to reopen. This is unfounded and ignores the timing of Plaintiffs’ own requests and the Court’s orders. Plaintiffs’ first wave of requests, which were served by April, were relatively narrow. Meta conducted a reasonable search, collected from potentially relevant custodians, and then reviewed and responded to those specific requests in advance of the July substantial document completion deadline and the original September 30 fact discovery cut-off. Plaintiffs then sought to extend discovery and Meta responded to a second wave of broader RFPs over the last few months: September 30 (Meta Resp. to Pltfs’ RFPs 76-78, 4th Set); November 8 (Meta Resp. to Pltfs’ RFPs 79-130, 5th Set); November 18 (Meta Resp. to Pltfs’ RFPs 131-136, 6th set). On October 4, the Court also granted Plaintiffs’ request to add five (5) ESI custodians and ordered Meta to produce documents from those custodians. (Dkt. 212). As Meta explained, this process took over 8 weeks to complete for its original custodians (Dkt. 190 at 8; Dkt. 190-8 ¶¶ 2-4). Plaintiffs never challenged that production timetable. Meta’s production over the last several weeks, culminating in its December 13 production, reflects those custodial searches and its updated productions in response to Plaintiffs’ 4th through 6th set of RFPs.

Meta’s position

C. Other Discovery Disputes

Leave a Reply


Discover more from Chat GPT Is Eating the World

Subscribe now to keep reading and get access to the full archive.

Continue reading