Product1 publisher3 min readPublished
OpenAI and Microsoft say plaintiffs' lawyers paid for research to supply evidence their clients lacked
OpenAI and Microsoft accuse Susman Godfrey, the plaintiffs' firm in the consolidated AI copyright suit, of paying for research and hiding the payment. At stake is the kind of market-dilution evidence a federal judge said last year that authors suing Meta had not presented.
The Product Desk · Product desk
What happened
- OpenAI and Microsoft accuse plaintiffs' firm Susman Godfrey of paying for research its clients lacked, hiding the payment, and bringing the work in outside the normal expert process.
- An arXiv preprint earlier this year, led by Tuhin Chakrabarty with copyright scholar Jane Ginsburg among its four authors, claimed generative AI floods and dilutes the book market.
- Publishing analyst Thad McIlroy faulted the paper for relying on Kindle Unlimited to get copies of the books it analysed.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- constraint If the court accepts the defendants' account, the plaintiffs could lose the study Techdirt identifies as their answer on market dilution.
- exposure Research brought into AI copyright suits now invites scrutiny of who paid for it and whether the opposing side was told.
- precedent Chhabria's opening for dilution claims survives this motion either way, so the next book plaintiffs have reason to bring that evidence through the expert process from the start.
A product team wants to ship a feature built on OpenAI or Microsoft models. Whoever signs off on it reads AI copyright news for one answer: did the risk go up or down? This filing is about how one set of research reached the court [1]. It affects one argument, market dilution, in a New York Times suit that has been folded into a much larger class action [3].
Dilution is where plaintiffs have had the least evidence. According to Techdirt, two federal judges in California split last year. William Alsup found AI training to be fair use in the Anthropic case, and Vince Chhabria found against fair use in the Meta case [5]. Alsup did not accept the dilution claims [13]. Chhabria called dilution "the potentially winning argument," then wrote that "the plaintiffs barely give this issue lip service, and they present no evidence about how the current or expected outputs from Meta's models would dilute the market for their own works" [6].
Techdirt's account presents the arXiv preprint as the dilution evidence at the centre of the dispute [7]. Chakrabarty earned his PhD at Columbia in 2024, where Ginsburg teaches, and now teaches computer science at SUNY Stony Brook [8]. McIlroy, who runs The Future of Publishing, objected to how the study got its books [9]. "But the author earns income on KU solely on the number of actual pages of their book that are read by a subscriber. Just getting downloaded provides no income," he wrote [10].
Here is what gets pitched. Ed Lee, who tracks AI lawsuits at ChatGPT is Eating the World, called the filing an "explosive motion" [4]. Techdirt called it a "massive bombshell (if not fraud on the court)" [11]. The same outlet had called the recent fuss over offhand employee remarks "something of a nothingburger," because those remarks should not affect the fair-use questions [12]. Here is what is being done: an evidence fight over whether paid work got around the rules for expert witnesses [1]. The reporting does not include a payment amount or a ruling.
Even if OpenAI and Microsoft win this motion, the fair-use question stays open. Liability for companies building products on these models is a further step away than anything in the filing or the coverage.
I'd sort each AI copyright headline on two questions. Does it change the legal rule, or only the evidence in one case? Does it come from the case your vendor is defending, or from another court? The box that justifies reopening a rollout is a ruling on the rule in your vendor's case. This motion sits in the box beside it: evidence, in your vendor's case, so it goes on a watch list. Rulings on the rule in other courts, like the Alsup and Chhabria split, tell you the law is unsettled [5]. Evidence fights in other courts can wait.
On this motion, I'd leave Monday's plans alone. The cost of waiting comes due only if the court lets the study stand. Then dilution becomes an argument with evidence behind it in the case OpenAI and Microsoft are defending, and the item moves up a box.
What to watch
- Whether the judge strikes the dilution research or orders Susman Godfrey to disclose what it paid and to whom.
- Susman Godfrey's response to the motion, and whether it disputes that the research was paid for.
- Whether plaintiffs in other AI book cases keep citing the dilution preprint or drop it.