Context: Several copyright infringement actions by newspaper publishers and other plaintiffs from the news industry are pending, and have been consolidated, in the Southern District of New York. The highest-profile one of those cases was brought by the New York Times (December 27, 2023 ai fray article). Other cases were brought by, for example, a group of other publishers led by the Daily News (April 30, 2024 ai fray article). In June 2024, the Center for Investigating Reporting (CIR) filed the third case to become part of this set of consolidated matters (BakerHostetler article). OpenAI (February 9. 2024 ai fray article) and Microsoft brought motions to dismiss parts of the NYT’s case, and subsequently also with respect to the correspondings parts of the related New York cases.
What’s new: Judge Sidney H. Stein of the United States District Court for the Southern District of New York yesterday adjudicated (PDF) the motions to dismiss further to a mid-January hearing. Judge Stein granted certain parts of them (common law unfair competition by misappropriation, the “abridgment” claims by CIR against OpenAI, all Digital Millennium Copyright Act (DMCA) claims against Microsoft and the bulk of the DMCA claims against OpenAI) and denied others (OpenAI’s challenge to direct infringement claims going back more than three years from the filing of the complaints, Open AI’s and Microsoft’s motions to dimiss the contributory copyright infringement claims, and both defendants’ motions to dismiss the trademark dilution claims the Daily News brought under federal and state law; and of the DMCA part, just the removal of copyright management information (CMI) claims against OpenAI in the Daily News and CIR cases). The reasons will be stated in a follow-up order in due course.
Direct impact: This is a partial win, of limited scope but undoubtedly of significant value, for Microsoft and OpenAI. It shows that the plaintiffs were experimenting with overly aggressive theories. To the extent that media reports place the emphasis on the judge allowing the case to go forward, they miss three key points. First, the defendants were never seeking the dismissal of those complaints in their entirety, so it was clear even prior to yesterday’s decision that the consolidated litigations would go on in some way, shape or form. The core of the cases was never at risk. Second, it’s the maximum penalty for plaintiffs to see any of their claims thrown out at the motion-to-dismiss stage (where the hurdle for claims to survive, which amounts to mere plausibility, is very low). Third, the fact that not all challenged claims were dismissed still does not mean that there will be a trial of any meaningful scope: defendants’ lawyers made it clear from the beginning that they would seek a dismissal on summary judgment, based on the fair use defense. The hurdle would be high for an interlocutory appeal of Judge Stein’s order to the United States Court of Appeals for the Second Circuit.
Wider ramifications: There are other AI copyright cases pending in other parts of the United States, the second major hotspot being the Northern District of California. Judge Stein’s decision is not binding there, and even a Second Circuit appellate decision wouldn’t be. Also, there are fact-specific reasons for which cases brought by, for instance, book authors are distinguishable from newspaper publishers’ complaints. One can see, however, a trend toward dismissing DMCA claims, though Judge Stein permitted more than the court in California (July 9, 2024 ai fray article). The United States Court of Appeals for the Ninth Circuit will presumably rule on that topic later this year (otherwise in 2026), but that decision will not be binding in the Second Circuit (to which New York belongs).
For defendants, any claims they can defeat at the motion-to-dismiss stage are a bonus. For plaintiffs, they are not necessarily a disgrace as lawyers have to try claims that are not only slam dunks, but dismissals definitely show that someone went too far in the opinion of the trial judge.
If the entire case had been thrown out (which, again, was not even on the agenda), the defendants could appeal now. But there still is a case left, albeit a narrower one than before.
A party could theoretically attempt an interlocutory appeal, but would find it hard to persuade the appeals court that there should be an exception from the rule that appeals should be brought from a final judgment. An interlocutory appeal in a DMCA context was allowed in the Northern District of California (September 28, 2024 ai fray article) and a New York judge may consider it desirable to enable the Second Circuit to speak out as well. If both those circuits agreed on DMCA claims in the AI copyright context, other circuits (with fewer such cases anyway) would be fairly likely to follow suit, and if not, there would be a circuit split warranting Supreme Court review.
There actually is another news-related case in the Southern District of New York that was not merged into the NYT-Daily News-CIR trio of cases: Raw Story Media v. OpenAI. It’s about news sources as opposed to newspapers, and it’s all about the DMCA. It was dismissed last year (November 8, 2024 ai fray article) without prejudice, meaning that an amended complaint can be submitted, but with skepticism as to whether that would change the outcome. Judge Colleen McMahon has yet to decide on whether to grant leave to amend the complaint. If she denies leave, or if she grants it but a renewed motion to dismiss succeeds, then there will be a final judgment, thus another (though arguably inferior) route for AI-related DMCA claims to the Second Circuit.
Assuming, for the purposes of discussion, that Judge Stein’s order in the NYT-Daily News-CIR cases will stand, the relatively most valuable win for the plaintiffs is the survival of the contributory infringement claims, but the dismissal of their (somewhat daring) unfair competition (by misappropriation) claims will hurt them because it’s a theory that they could have used, especially in psychological terms, to influence the jury’s stance on damages (should the case reach the remedies stage).
Judge Stein’s forthcoming reasons will be instructive, but it was obvious from the outset of these cases that the single most important juncture of the pretrial proceedings would be summary judgment. As the Supreme Court noted (and practiced) in Oracle v. Google, fair use is decided by judges: it involves facts, which would normally be left to a jury, but at its core it constitutes an equitable balancing act. If the defendants prevail on fair use, there will be no more copyright infringement claims left for a jury to decide. Combined with what the court has already thrown out at the motion-to-dismiss stage, those cases would be reduced to rubble. Otherwise there would be a trial with a multi-billion-dollar damages claim unless the court brings down the numbers to a more moderate level, using its gatekeeper powers under Daubert.
The motions to dismiss have been reasonably successful. They could have been even more successful, but given the low hurdle, the plaintiffs presumably hoped they could fend off the motions to dismiss to a greater extent.