When News Became Training Data: What the Unsealed Filings Say

Unsealed court papers show Microsoft and OpenAI employees worried about how A.I. systems use news stories. The New York Times and other publishers filed the documents in a copyright case.

A Microsoft scientist, Brent Hecht, wrote that the scraping of news was the largest theft of labor in human history. The filings say OpenAI and Microsoft used millions of news articles to train models.

Microsoft data shows its Copilot answer engine cut click-through rates to The New York Times by as much as 93%. The filings also say staff discussed ways to share training data in projects called Project Mango and Project Taxi.