Executives from Microsoft and OpenAI have made rare admissions in court testimony that their AI products are substituting for news publisher content, posing a genuine threat to the journalism industry.

Microsoft CEO Satya Nadella testified that conversations with chatbots have "replaced" user visits to publisher websites, while OpenAI senior executive Nick Turley wrote more bluntly, "Our products are largely substitutive, full stop."

These revelations come from a partially unsealed motion for summary judgment filed by news publishers on Thursday, as part of a multi-party lawsuit alleging that OpenAI and Microsoft used news articles without authorization to train AI models. The Trump administration has weighed in on the case, urging the court to rule that training AI on copyrighted works constitutes "fair use."

According to Bloomberg, when OpenAI employees informed company president Greg Brockman about methods to circumvent The New York Times' paywall to assist with data scraping, his response was, "Ah, nice."

Where the evidence leads

This unsealed motion is a follow-up to redacted versions filed roughly two weeks ago, directly exposing how OpenAI and Microsoft executives perceive the nature of their AI products.

Nadella stated in his testimony that content behind paywalls should be licensed for AI training purposes. He added that had he known OpenAI had scraped and trained on paywalled content, he would have exercised "Microsoft's right" to demand the company retrain its models.

Dr. Brent Hecht, director of applied science at Microsoft, indicated that copying millions of news articles without permission could amount to "the largest theft of labor in human history." He further argued that if OpenAI and Microsoft won on a fair use defense, it would "completely mock the very concept of fair use."

OpenAI's policy director Jack Clark also wrote internally that the company's work "will increasingly lead us to create systems that replace the labor of people who define the 'culture' of society."