One judge now oversees a dozen AI copyright lawsuits

The writer's problem

By early 2025, authors, news publishers and a YouTube creators' class had filed a dozen separate copyright suits against OpenAI and Microsoft in two different federal courts, each proceeding on its own schedule with overlapping discovery about the same training process. The Judicial Panel on Multidistrict Litigation's own transfer order, entered on its docket on 3 April 2025, describes the practical problem this created: duplicative depositions, competing rulings on the same factual questions, and courts unable to coordinate class certification across cases that shared discovery but not a docket.

What the documents show

The order states the panel centralized all twelve actions, listed on its own schedule and including the Authors Guild's suit, the Times's suit, and cases brought by Sarah Silverman and other authors, in the Southern District of New York before Judge Sidney Stein. It records that OpenAI itself moved for centralization in California, that most plaintiffs opposed centralization outright or proposed a different venue, and that the panel chose New York because eight of the twelve suits already sat there and Stein already presided over six of them. The order is explicit that consolidation addresses shared factual questions about how the models were trained, not the differences in legal theory among the cases: it separately notes that the news plaintiffs allege both training-data use and verbatim output reproduction, that a Digital Millennium Copyright Act claim from two outlets does not even allege copyright infringement, and that a YouTube-creator suit involves video transcripts rather than books. The panel states these differences do not block consolidation because a transferee court can manage separate tracks for different claims.

The editorial choice

A publisher or writer following any one of these suits should track the transferred case number and judge, since procedural developments, like the September 2026 summary-judgment motions, now apply across the consolidated docket rather than to a single plaintiff's case alone. Treating consolidation as a merger of legal theories, rather than a shared pretrial schedule, is a common and avoidable misreading of what the order does.

What stays with the author

Each plaintiff group keeps its own distinct claim and remedy theory inside the consolidated docket; the order does not merge a book-training claim with a verbatim-reproduction claim into one cause of action. What remains unresolved is how the transferee court will ultimately separate these tracks for trial, if the case reaches one.

• Which of the twelve original suits does a given report actually concern, and is it still identified by its old docket number?
• Does a claim about this litigation's status reflect the consolidated MDL, or an earlier, since-transferred filing?
• Are training-data and output-reproduction theories being described separately, as the panel's own order does?

Consolidation changed where and before whom these cases proceed, not what each set of plaintiffs must still separately prove.