NYT-led publishers file summary judgment brief citing internal OpenAI/Microsoft messages calling AI training "astonishing theft" and admitting chatbots substitute for journalism

A group of publishers led by The New York Times, alongside Ziff Davis (CNET, IGN, PCMag), the Daily News group (Chicago Tribune, Denver Post), the Center for Investigative Reporting (Mother Jones) and The Intercept, filed a 92-page joint summary judgment brief against OpenAI and Microsoft in the consolidated copyright MDL in US District Court, New York, seeking billions in damages. The brief quotes internal emails, Slack messages and sworn deposition testimony from named executives, including Microsoft's Brent Hecht, OpenAI's Nick Turley, Satya Nadella and Greg Brockman, and alleges paywall bypassing, misuse of a licensed NYT corpus outside its terms, and a filter built after being sued to suppress plaintiff-sourced outputs. This account comes secondhand via The Decoder, with no direct link to the filing itself.

This is a filed legal document with named deponents and discovery material, not a press release or opinion piece, so it carries real evidentiary weight in a case that will help set the legal boundaries for training AI on copyrighted text. If the internal admissions hold up, they attack all four fair use factors at once, which is the crux of nearly every AI copyright case in the US right now, not just this one. Enterprises building on OpenAI/Microsoft models, publishers weighing licensing deals, and investors pricing litigation risk all have a direct stake in how a judge treats "we knew it was substitutive" language from the companies' own people.

"Astonishing theft" is a quote plaintiffs' lawyers chose to headline with, not a neutral description, and it appears in a brief written to win, so treat it as advocacy framing rather than settled fact until OpenAI/Microsoft respond. The claim that this could "reshape how AI training is treated legally" is plausible but speculative; summary judgment briefs are argued positions, and OpenAI and Microsoft have not yet filed their opposition or a competing account of the same internal messages. No links to the actual filing, deposition transcripts or docket were provided, so none of this can be independently verified from the sourcing given here.

1. 01OpenAI and Microsoft's opposition brief and their characterisation of the same internal messages, expected on the standard MDL briefing schedule
2. 02Whether the presiding judge rules on summary judgment or sends the fair use question to trial, and any explicit findings on the four statutory factors
3. 03Independent confirmation of the quoted Copilot click-through drop figures (51-94% by publisher) and the paywall-bypass allegations from court filings or other outlets covering the docket directly