IP Law Daily, COPYRIGHT NEWS: JPML consolidates OpenAI copyright cases in New York federal court, (Apr 7, 2025)

By George Basharis, J.D.

Multiple copyright lawsuits against OpenAI and Microsoft will be heard together in Manhattan, despite objections from several plaintiffs.

The United States Judicial Panel on Multidistrict Litigation (JPML) has ordered the consolidation of 12 copyright infringement lawsuits against OpenAI and Microsoft under a single federal judge in New York. The decision centralizes cases from both California and New York in the Southern District of New York under Judge Sidney H. Stein, an experienced transferee judge who already presides over six of the eight actions in that district (In Re: OpenAI, Inc., Copyright Infringement Litigation, No. 3143 (J.P.M.L. Apr. 3, 2025)).

The 12 actions listed in the order include four from the Northern District of California and eight from the Southern District of New York. These cases represent various plaintiff groups including authors, publishers, news organizations, and content creators, all alleging different forms of copyright violations related to AI training and outputs.

While most plaintiffs focus on the unauthorized use of their content as training data, news publishers also claim the AI systems reproduce their content in outputs, and certain media organizations allege violations of the Digital Millennium Copyright Act (DMCA) regarding removal of copyright management information.

Panel orders centralization despite opposition. The OpenAI defendants—comprised of 11 related OpenAI entities—originally moved to centralize the litigation in the Northern District of California under a procedure established by federal law in 28 U.S.C. § 1407, which allows related cases to be consolidated for pretrial proceedings. Microsoft Corporation, a defendant in several New York actions, supported centralization in either California or New York and requested an experienced transferee judge.

While plaintiffs in a case filed in the Northern District of California (Millette) by a specific category of content creators (YouTube video makers) supported consolidation in California, most responding plaintiffs opposed centralization entirely. However, they did offer alternative positions if consolidation was ordered.

Plaintiffs in California class actions alternatively supported the Northern District of California, while plaintiffs in New York class actions preferred the Southern District of New York. These actions were brought by authors of fiction and non-fiction works.

A third group—plaintiffs in the New York “News” cases (New York Times, Daily News, and Center for Investigative Reporting) and the New York “DMCA” cases (Raw Story Media and The Intercept)—alternatively suggested that if an MDL were created, their specific cases should be centralized in New York.

Common questions of fact prevail. Despite the opposition, the judicial panel determined the cases share common factual questions arising from allegations that OpenAI and Microsoft used copyrighted works without consent or compensation to train their large language models (LLMs), including GPT-4, which power generative AI products like ChatGPT and Microsoft’s Copilot (formerly Bing Chat).

The panel noted that each action “will involve overlapping, complex, and voluminous discovery regarding how defendants trained and designed their [large language models, or LLMs].” Given the “novel and complicated nature of the technology,” the panel anticipated overlapping experts across these actions.

The order emphasized that centralization would “eliminate duplicative discovery, prevent inconsistent pretrial rulings (particularly as to class certification), and conserve the resources of the parties, their counsel, and the judiciary.”

Different claims present no obstacle. The opposing plaintiffs emphasized the distinct nature of claims across the consolidated cases. Authors of fiction and non-fiction works allege copyright infringement through unauthorized use of their novels and books as training data. News organizations like The New York Times claim dual infringement—both during training and in the AI’s ability to generate “verbatim and detailed summaries” of their news content.

Media outlets including Raw Story Media and The Intercept base their claims on the DMCA, focusing on the removal of copyright management information rather than direct copyright infringement. Meanwhile, the Millette action represents YouTube content creators whose video transcripts were allegedly used without permission.

Despite these variations, the panel ruled that the common questions regarding how OpenAI and Microsoft trained their language models outweighed the differences. The order indicated that the transferee court could establish separate tracks for different claim categories to ensure efficient case management.

Advanced proceedings not a barrier. Opposing plaintiffs also argued that pretrial proceedings in the class actions and News actions were too advanced for centralization to be beneficial. The panel was not persuaded, noting that counsel stated during oral argument that only four depositions had been taken so far, with at least forty more needed in the class actions alone.

The panel highlighted that the parties had been unable to agree on a deposition coordination protocol to eliminate duplicative depositions of defense witnesses. The order noted that “much of the recent briefing by the parties and oral argument before the Panel consisted of counsel blaming the other side for the breakdown in negotiations.”

According to the panel, “Opportunities abound for the transferee judge to redirect the parties onto a more efficient path and eliminate the potential for duplicative discovery and pretrial motion practice, as well as inconsistent pretrial rulings and scheduling.”

Stipulation does not prevent centralization. Finally, opposing plaintiffs argued that OpenAI could not seek centralization under the terms of a January 2024 stipulation and order in the New York class actions. The panel declined to resolve this dispute, stating that "regardless of whether OpenAI could request centralization under the stipulation, this Panel itself can certainly order it" under its statutory authority.

The panel cited 28 U.S.C. § 1407(c), which grants it authority to centralize civil actions upon its own initiative and referenced a recent Ninth Circuit decision (Uber Techs., Inc. v. U.S. Jud. Panel on Multidistrict Litig., No. 23-3445, __ F.4th __, (9th Cir. Mar. 10, 2025), which held that the statute “does not create an individual right to centralization that may be waived but instead vests the JPML with a power to manage the federal docket by centralizing cases that is unfettered by private agreements.”

New York selected as transferee district. The Southern District of New York was deemed an appropriate transferee district for this litigation, as eight of the twelve actions are already pending there. Several plaintiffs, as well as Microsoft, alternatively suggested or did not oppose centralization in this “convenient and accessible venue.”

Judge Sidney H. Stein, described as an experienced transferee judge, already presides over six of the eight actions in the district. The panel noted that both he and Magistrate Judge Ona T. Wang “have devoted substantial time and resources to this litigation, and thus are well situated to steer this litigation on a prudent and expeditious course.”

Technical understanding of LLMs. The transfer order includes a detailed explanation of how large language models work, noting they are “algorithms that predict words that are likely to follow a given string of text based on the potentially billions of examples used to train it.”

According to the parties, LLMs encode information as “parameters,” with approximately 1.76 trillion parameters in the GPT-4 model. Training involves storing encoded copies of works in computer memory, repeatedly passing them through the model with words masked out and adjusting parameters to minimize the difference between masked-out words and predictions. The panel noted that defendants had “already produced petabytes of data from their training sets,” highlighting the massive scale of discovery involved in these cases.

The Case is MDL No. 3143.

Judge: Stein, S.

Attorneys: Cadio R. Zirpoli (Joseph Saveri Law Firm, LLP) for Paul Tremblay. Andrew Michael Gass (Latham & Watkins LLP) for OpenAI, Inc.

Companies: OpenAI, Inc.

News: Copyright TechnologyInternet GCNNews AINews