Context: The New York Times is one of the publishing companies that prefer pursuing copyright claims against AI providers over license deals (May 5, 2024 ai fray article). In February, OpenAI filed a motion to dismiss, not with respect to the entire NYT case but certain parts, particularly in order to narrow discovery (February 29, 2024 ai fray article). One of the arguments is that the NYT’s infringement claims are barred to the extent they go back more than three years from the late December 2023 filing of the complaint (which relates to GPT-2 and GPT-3). The NYT obviously opposed the motion (March 17, 2024 ai fray article).
What’s new: Yesterday (May 9, 2024), the NYT’s lawyers informed Judge Sidney Stein (United States District Court for the Southern District of New York) of a Supreme Court decision that potentially supports the NYT’s damages period. But it doesn’t necessarily do so, and in any event it is unrelated to the merits of the copyright infringement and other claims in the case.
Direct impact: The defendants (Microsoft and OpenAI) could ask the court for permission to file a response to the NYT’s request for judicial notice. It’s also worth noting that the NYT’s lawyers had argued that even a damages period limited to three years would cover the economically most important ones of the accused actions. Those older GPT versions are less important according to what they plausibly wrote.
Wider ramifications: The question of how long the damages period may be is also relevant to the more recently-filed case by various newspapers belonging to hedge fund Alden Capital (April 30, 2024 ai fray article), a case that will likely be consolidated with the NYT’s lawsuit.
Here’s the NYT’s lawyers’ letter to the court:
The Supreme Court decided Warner Chappell v. Nealy with a 6-3 majority. It was not a split among partisan lines: the per curiam, authored by Justice Kagan, has the support of three Democratic and three Republican appointees. All three dissenters are Republicans (Justice Gorsuch authored the dissent, and Justices Thomas and Alito joined).
In that case, the plaintiff filed an infringement action in 2018 seeking copyright damages going back to 2008. In other words, the claims went back by a decade. The question was then whether that period was partly time-barred by under ยง507(b) of the Copyright Act, according to which a lawsuit must be brought “within three years after the claim accrued.” The first question is then whether “accrued” means that an infringing act is committed or whether the discovery rule applies, in which case the right holder either was aware or a diligent right holder would have been aware of the infringement.
Another question came up in that Warner Chappel case: even if one assumed that the discovery rule applied, could there be a limit to a three-year damages term anyway? That’s what the district court in that case thought, but it allowed an interlocutory (prior to final judgment) appeal. The Eleventh Circuit then had to decide between either taking the same restrictive position as the Second Circuit (which is where the NYT’s case is being litigated) or the more permissive one of the Ninth Circuit, according to which there is no limitation as long as the lawsuit is brought on a timely basis. The Eleventh Circuit opted for the more permissive stance, and the Supreme Court majority has now affirmed that one.
But the Supreme Court did not resolve the question of whether the discovery rule applies. The opinion makes it clear that this question was not properly before the SCOTUS (and it may actually be addressed soon in another case). So the Supreme Court assumed, without endorsing it, the applicability of the discovery rule, just because that’s what the appealed decision (the one the Eleventh Circuit had made) was based on the same assumption.
At first sight, it may appear counterintuitive. It looks like the second step was made before the first. The first is whether the discovery rule applies, and the second is whether (if it applies) the damages period is limited to three years if the lawsuit is brought on a timely basis. The majority’s answer to the second question was no.
The dissent, however, essentially says “first things first”: the three Republican dissenters think there was no point in answering the second question without addressing the first one in a first step, particularly because they believe “the [Copyright] Act almost certainly does not tolerate a discovery rule.”
So what does all of this mean now for the NYT’s case?
The NYT’s lawyers feel the SCOTUS decision in favor of a copyright holder may dissuade Judge Stein from narrowing the case. But the Supreme Court might still throw out the discovery rule in another case (as the minority would already have done on this occasion). And even if it applied, OpenAI points to an NYT article from 2020 that allegedly showed they had knowledge of certain alleged infringements, or at least enough of a basis that a diligent right holder would have taken a closer look. The NYT’s lawyers argue the article falls far short of that. So that’s another disputed question.
The NYT’s “extensive” reporting on its own case is also the subject of a discovery dispute (May 4, 2024 ai fray article).
On the bottom line, yesterday’s Supreme Court decision may help the NYT with a view to the damages period, or it may not. There still are a couple of question marks.
Finally, here’s the NYT’s letter to the court: