Microsoft Executive Called AI Training "Largest Theft of Labor in Human History," Which the Company Notes Is Also a Record
Unsealed filings show the remark was made privately, which sources say is where Microsoft keeps its most accurate statements.
REDMOND, Wash. — Newly unredacted filings in The New York Times's three-year-old copyright lawsuit reveal that a senior Microsoft executive privately described the AI industry's training practices as "the largest theft of labor in human history," a phrase the company said Wednesday had been taken out of context.
"The context was a private conversation in which we were being candid," a company spokesperson said. "We would ask that it be judged by the standards of our public statements, which are different."
The filing, which draws on the newspaper's brief rather than the still-sealed exhibits, also quotes OpenAI leadership describing its models as an "existential threat" to the publishers whose work trained them, an assessment the company's fair-use defense characterizes as "a compliment."
"Fair use requires that the use not harm the market for the original," said Deshawn Whitfield, a professor of copyright law. "Their own executive called it the largest theft in history. As a legal posture, that is pleading not guilty while holding the television."
Microsoft argued in a response that the remark was "hyperbole," a form of speech the company said it does not typically train on. It added that the executive had not said "theft" in the legal sense but in "the sense people use when something is very good."
"The thing is, he was proud," said a person familiar with the conversation. "You do not say 'largest in human history' about something you are ashamed of. You say 'a rounding error.'"
The Trump administration filed a brief earlier this month in support of the companies' position, which the companies said they had read, in full, without permission.
At press time, Copilot had summarized the filing as "generally favorable."