What came out on Thursday
A largely unredacted version of The New York Times’s summary judgment brief against OpenAI and Microsoft was unsealed on 17 September in the Southern District of New York. It is one of a set of competing motions — OpenAI filed its own memorandum on 4 September — and the ruling on them can dispose of some claims or narrow what a jury would be asked to decide.
The brief is significant less for its legal argument than for what it puts on the public record: internal messages from inside both defendants, written years before the case, that the Times is using to argue the companies knew what they were doing.
The Microsoft memo
The quotation that has travelled furthest comes from Brent Hecht, Microsoft’s director of applied science. In a January 2023 internal memo, reported by TechCrunch and others from the unsealed filing, he described the practice as an astonishing theft of unprecedented proportions, and as the largest theft of labour in human history.
It is one employee’s view in one memo, not a corporate position, and Microsoft has not said it was company policy. But it is the kind of document that is hard for a defendant to explain away, because it shows the question was asked inside the building and answered in those terms.
The OpenAI messages
The brief also quotes Nick Turley, who runs ChatGPT, describing the product as an existential threat to publishers and as largely substitutive of their work, with the substitution expected to increase as the models improve. OpenAI president Greg Brockman is quoted saying the models are excellent at news.
Those are the sentences the Times needs. A fair-use defence turns substantially on market effect — whether the new work displaces demand for the original — and internal statements that the product substitutes for publishers cut directly against the companies’ position.
The scale alleged
On the Times’s account, content was scraped from more than 10 million articles, with close to a third of those coming from the Times alone. The brief also describes paywall circumvention and the removal of copyright management information from training data — the latter a separate claim under US law from infringement itself.
None of that has been tested at trial. These are the plaintiff’s characterisations of documents produced in discovery, filed in support of its own motion.
Why this case in particular
The Times sued OpenAI and Microsoft in December 2023. It is the furthest advanced of the major news-publisher cases, it involves the largest AI company and its largest investor together, and the summary judgment stage is where a fair-use ruling would first be made rather than argued.
A decision for the Times on liability would reset the licensing negotiations every publisher is having. A decision for OpenAI on fair use would do the same in the opposite direction.
What to watch
The court’s ruling on the competing motions, and whether it addresses fair use squarely or defers it to trial. The Authors Guild’s parallel suit had briefs unsealed alongside the Times’s, so both are now readable in full.
