Unsealed New York Times brief exposes internal AI-substitution warnings from Microsoft and OpenAI

• Microsoft director of applied science Brent Hecht described large-scale scraping as “an astonishing theft of unprecedented proportions” and possibly “the largest theft of labor in human history,” according to the plaintiffs’ brief. ^[1]
• OpenAI head of ChatGPT Nick Turley wrote that the company’s products were “largely substitutive” and that publishers faced an “existential threat,” while Microsoft CEO Satya Nadella testified that chatbot conversations had “substituted” for publisher sites. ^[2]
• The statements appear in a plaintiff-authored brief that summarizes discovery evidence; some underlying exhibits remain sealed, and the quotations are not judicial findings. ^[3]
• The case is at the summary-judgment stage. A ruling could resolve some claims or determine which issues proceed to trial. ^[4]

A partly unredacted court brief has revealed internal warnings from Microsoft and OpenAI employees that AI systems could replace publisher websites and weaken the economic foundation of journalism, giving The New York Times fresh evidence in its copyright case against the companies. [1]

In the news plaintiffs’ summary-judgment brief, Microsoft director of applied science Brent Hecht is quoted describing the mass collection of online material for AI training as “an astonishing theft of unprecedented proportions” and possibly “the largest theft of labor in human history.” [1]

The qualification matters. The public filing presents Hecht’s language as a warning about how “millions of people” might view companies “hoovering up” their work; it does not show him making a legal conclusion that Microsoft infringed copyright. The underlying document containing the exchange remains sealed. [3]

The filing also quotes OpenAI head of ChatGPT Nick Turley writing that “our products are largely substitutive, period,” and warning that publishers faced an “existential threat.” Microsoft CEO Satya Nadella, in deposition testimony cited by the plaintiffs, said chatbot conversations had “substituted” for publisher sites. [2]

What the filing supports #

The Times uses the statements to argue that Microsoft and OpenAI understood their products could function as substitutes for the works used to train them. That argument bears on fair use, including the purpose and character of the copying and its effect on the market for the original works. [1]

The brief also describes a Microsoft document warning of a “doom loop” in which AI-generated answers reduce traffic to the websites that supply the systems’ information. The plaintiffs present that concern as evidence that the companies recognized a conflict between their products and the publishers whose work helped train them. [1]

The record does not establish that every statement reflected a company-wide view. Hecht was a Microsoft director of applied science, while Turley was an OpenAI executive; the filing attributes the statements to particular people and documents rather than presenting them as formal corporate admissions. [3]

Responses from Microsoft, OpenAI and the Times #

Microsoft spokesman Alex Haurek said Hecht’s writings reflected “one employee’s individual perspective,” were not legal analysis and did not represent Microsoft’s views. He said Nadella’s testimony and Microsoft’s litigation position were consistent, describing the testimony as an observation about changes in how people find and consume information rather than a conclusion about copyright law. [5]

Microsoft’s court position is that using copyrighted material to train AI systems can qualify as transformative fair use and that Copilot is not a substitute for publishers’ journalism. [5]

OpenAI did not return requests for comment in the independent coverage reviewed. Its public litigation position is that AI training is protected by fair use and that the Times’ claims have narrowed as the case has progressed. OpenAI says it filed its summary-judgment memorandum concerning the Times on September 4, 2026. [6]

The Times’ position is set out in the brief: the companies copied millions of articles without authorization, used the material to build commercial systems and created outputs that can reproduce or substitute for original reporting. The plaintiffs are asking the judge to resolve at least some liability issues without a trial. [1]

The case is narrower than the headlines suggest #

The Times sued Microsoft and OpenAI on December 27, 2023, alleging that the companies used its journalism in AI systems without permission. The litigation was later consolidated with related copyright cases involving other news organizations and authors in multidistrict proceedings before the same federal judge. [7]

The court has already narrowed some claims. In an August 6, 2026 order, Judge Sidney Stein dismissed with prejudice the news plaintiffs’ contributory-infringement claims based on a “material contribution” theory and dismissed trademark-dilution claims. The judge rejected a request to add different contributory theories against Microsoft after finding that the plaintiffs had waited too long to amend their pleadings. [4]

Those rulings did not resolve the core dispute over alleged direct infringement through training or outputs, or whether the defendants’ uses qualify as fair use. The parties filed their initial summary-judgment motions on September 4, and a less-redacted version of the news plaintiffs’ filing became public on September 17. [6]

A summary-judgment ruling could dispose of some claims or define the issues for trial. The newly public statements may help the Times argue that the companies’ internal assessments conflict with their litigation positions, but the statements alone do not establish copyright infringement. That determination remains with the court. [3]

Companies mentioned #

Further sources #

The stories that matter, in one email. Free — unsubscribe anytime.