Unsealed Court Papers Show a Microsoft Executive Privately Called AI Training on News Articles 'the Largest Theft of Labor in Human History'

The New York Times and four other publishers filed a public version of their argument for winning their copyright case against OpenAI and Microsoft without a trial, and it quotes the companies' own internal documents: a Microsoft scientist calling the practice "an astonishing theft of unprecedented proportions," and the head of ChatGPT writing that publishers face an "existential threat." Microsoft says those were one employee's opinions. The companies argue their use of articles is legal, and the U.S. government sided with them this month.

In December 2023, The New York Times sued OpenAI and Microsoft for using millions of its articles, without permission, to build their chatbots. Today the newspaper, together with the New York Daily News, the Center for Investigative Reporting, The Intercept and the publisher Ziff Davis, filed a public version of a 92-page brief asking the judge to rule in their favor on the main questions without a trial. Much of the underlying material had been sealed. The newly visible parts quote what the companies' own people wrote to each other.

The line that leads the brief comes from Microsoft's director of applied science, Brent Hecht, in a document the publishers say he wrote soon after the lawsuit was filed: "millions of people around the world will soon consider large models 'hoovering up' all their work to be an astonishing theft of unprecedented proportions." He added that "almost no one intended for content they created to be used in this fashion, nor are they compensated for its use." Elsewhere, the brief says, he called it perhaps "the largest theft of labor in human history." Another Microsoft document warns that "our AI content strategy has started a 'doom loop' that will hurt the performance of our models and the entire web at the same time: It is highly unusual that an end-product threatens the economic foundations of its essential suppliers," but that, it goes on, is the situation the company has created for its chatbot business. The brief cites Microsoft's own figures showing that when its Copilot chatbot answers a question instead of a traditional search, clicks through to the Times's and Daily News's websites fall by 83 to 93 percent.

On the OpenAI side, the brief quotes the head of ChatGPT, Nick Turley, writing that publishers face an "existential threat" from products that "are largely substitutive, period" and "will get more and more substitutive as they get better." Substitutive is the key word in copyright: it means the copy replaces the original rather than adding something new. The publishers say OpenAI understood this as early as June 2023. In another exchange, when a colleague told OpenAI's president, Greg Brockman, about "a hack to get around nytimes paywall," the subscription wall that keeps articles for paying readers, Brockman replied "ah nice." The brief also says OpenAI's training collections contain more than 91,692 copies of the publishers' articles and that OpenAI stripped out the copyright notices before training. And it quotes Microsoft's chief executive, Satya Nadella, testifying under oath that "anything that is paywalled should be licensed by anyone who wants to use it."

This is one side's brief, written by lawyers to persuade a judge, and the quotations were chosen for that purpose. Microsoft told the AFP news agency that Hecht's statements were "one employee's individual perspective" and "do not represent the company's views." OpenAI did not comment. The companies' defense throughout the case has been fair use, the rule in copyright law that allows using someone's work without permission when the new use transforms it into something different, and they argue that teaching an AI to write is exactly that. On September 1, the U.S. Department of Justice filed papers in the case urging the judge to agree, arguing that a ruling against the companies would weaken American competitiveness and national security. The Times answered that the government was siding with "trillion-dollar AI companies" at the expense of the people who make things.

What happens next is up to the judge, Sidney Stein, who will decide whether the questions can be settled on the paper record or need a jury. The reason today's filing matters to people outside the courtroom is simpler. For three years the argument has been about what the companies did. Now it is also about what they said to each other while they did it, and whether a business can call something legal in public and a theft in private, and be right both times.