Every week, the Array team reviews the latest news and analysis about the evolving field of eDiscovery to bring you the topics and trends you need to know. This week’s post covers the period of November 24-30. Here’s what’s happening.
New York Times v. OpenAI Highlights Discovery Challenges of AI Datasets
Copyright owners continue to do battle with generative AI companies over the use of their news articles, books, artwork, and more as training data, highlighting the complexity of eDiscovery involving large language models.
In its ongoing lawsuit against OpenAI and related parties in the U.S. District Court for the Southern District of New York, The New York Times recently filed a request that OpenAI, rather than the Times, should search its own databases to determine which of the newspaper’s articles were used as training data. And, if OpenAI would not conduct the requested searches, the letter further requested, it should be required to admit to which of the claimed copyrighted works were used as training data.
The discovery dispute arises from the structure set up to allow the Times to search OpenAI’s training data. OpenAI provided the Times with non-internet connected virtual computers – called the “sandbox” by the parties – to access and search OpenAI’s data. The Times alleges that OpenAI’s engineers accidentally deleted “all of News Plaintiffs’ programs and search result data stored on one of the dedicated virtual machines” and that the recreation of the folder structure and file names would require “significant person-hours and computer processing time.”
In response, OpenAI claimed that the plaintiffs themselves were responsible for the deletion and “Plaintiffs could re-run the searches to recreate the files with just a couple days of computing time.”
For its part, OpenAI claims that the Times requested a configuration change that caused the inadvertent deletion and that the Times also had been saving data on a hard drive that was intended only for OpenAI’s data. “Contrary to Plaintiffs’ insinuations, the facts are much more banal and the consequences too,” OpenAI’s attorneys wrote in a letter to Judge Ona Wong. “OpenAI did not delete any evidence.”
Writing for Ars Technica, reporter Ashley Belanger notes that, “This isn’t the only time that OpenAI has been called out for deleting data in a copyright case.” In a case in the Northern District of California brought by authors including comedian Sarah Silverman, the plaintiffs requested a Rule 30(b)(6) corporate deposition regarding datasets that the plaintiffs’ request noted “OpenAI now admits were deleted and witnesses knowledgeable about the creation of these datasets have apparently left the company.”
To belabor the obvious, large language models are, by nature, large. Moreover, the AI companies being accused of infringement often have a greater volume of relevant eDiscovery data than plaintiffs, as well as the technical know-how to sort through it. Litigation in which one side houses more documents than another is not new, but AI cases highlight just how much of a challenge it can be, particularly as parties, attorneys, and eDiscovery consultants tackle new technical issues.
Ninth Circuit Holds Rule 37(e) Precludes Inherent Court Authority for Deleted ESI Sanctions
The Ninth Circuit recently vacated a lower court’s sanctions for the defendants’ failure to preserve video surveillance footage in a civil case arising from an altercation between a probation officer and the mother of an individual under the officer’s supervision. The appellate court held that Federal Rule of Civil Procedure 37(e), rather than the inherent authority of the court, governed the sanctions award. This distinction made a difference in the case because Rule 37(e) has a more stringent intent standard than the court’s inherent authority.
On the basis of the higher intent standard, the Ninth Circuit reversed and remanded the verdict against the officer, whom it said had been hindered by the discovery sanction from presenting her defense.
In Gregory v. Montana, the plaintiff sued the state and the probation officer, alleging that the officer used excessive force in an encounter in the parking lot next to the probation office. Shortly after the incident, the plaintiff, Carrie Gregory had been charged with misdemeanor obstruction of a peace officer and her counsel in that matter called the office to request surveillance footage of the parking lot. The regional deputy chief allegedly assumed that the footage would be retained for 30 days, when in fact the automatic deletion period was after 17 days. After delays in sending the footage to Gregory’s defense counsel, the footage was eventually deleted in accordance with the 17-day policy.
Federal Rule 37(e) provides (emphasis added):
(e) Failure to Preserve Electronically Stored Information. If electronically stored information that should have been preserved in the anticipation or conduct of litigation is lost because a party failed to take reasonable steps to preserve it, and it cannot be restored or replaced through additional discovery, the court:
(1) upon finding prejudice to another party from loss of the information, may order measures no greater than necessary to cure the prejudice; or
(2) only upon finding that the party acted with the intent to deprive another party of the information’s use in the litigation may:
(A) presume that the lost information was unfavorable to the party;
(B) instruct the jury that it may or must presume the information was unfavorable to the party; or
(C) dismiss the action or enter a default judgment.
The Advisory Committee notes on the 2015 amendments state that Rule 37(e) “authorizes and specifies measures a court may employ if information that should have been preserved is lost, and specifies the findings necessary to justify these measures. It therefore forecloses reliance on inherent authority or state law to determine when certain measures should be used.”
The district court held that the state and the officer did not act with gross negligence or willfulness in failing to preserve the video footage, but found that they had acted with recklessness. On this basis and invoking its inherent authority, the court ordered issuance of a jury instruction that “it has been established as a matter of law that Defendant Williams used excessive force” against Gregory.
The Ninth Circuit held that the rule’s requirement of intent precluded the court from invoking its inherent authority and imposing severe sanctions for failure to preserve the video.
In a recent article on Law.com, H. Christopher Boehning and Daniel J. Toal of Paul, Weiss, Rifkind, Wharton & Garrison wrote that, since the 2015 amendments to Rule 37(e), there has been a split among courts, some of which continue to apply their inherent authority to issue sanctions for failure to preserve ESI. “[I]n finding that Rule 37(e) precludes a court from relying on its inherent authority to impose sanctions for the loss of ESI, the decision underscores the importance for practitioners to understand the possible boundaries of a court’s inherent authority in this area when seeking or defending against sanctions,” they wrote. “Notably, courts do not face the same limitations when imposing sanctions under other subsections of Rule 37, such as Rule 37(b) for the failure to comply with a court order.”
Other recent eDiscovery news and headlines:
- Florida’s Civil Procedure Rules Will Change How Trials Are Run (Bloomberg Law)
- What “Expenses” Can a Non-Party Recover for Complying with a Discovery Subpoena? (EDRM on JD Supra)
Julia Helmer; Director, Client Solutions
With 15 years of expertise, Julia excels at optimizing enterprise eDiscovery workflows from start to finish. With a deep understanding of how to seamlessly integrate workflows across various eDiscovery platforms, Julia creates tailored solutions for data identification, legal holds, ESI collections, and productions. By harnessing the power of Technology Assisted Review and Analytics, she delivers efficient, cost-effective results that align with best practices and budgetary constraints. Julia’s exceptional communication and customer service skills have fostered strong, lasting relationships with both clients and Project Management teams, enabling her to effectively problem-solve and drive success across numerous projects.
