Internal Docs Warn of Human Labor Theft in AI Model Training

Leaked court documents have revealed growing concerns within Microsoft and OpenAI over the use of millions of news articles to train AI models, labeling it "the biggest theft of human effort in history".

Details of the concerns

A Microsoft internal memo from 2023 shows staff fearing that reliance on journalistic content to develop massive language models could jeopardize the sustainability of the news industry. Brent Heist, director of applied science, warned that AI models could enter a vicious cycle: they depend on news content to improve, then compete with the very outlets that produce it, threatening the data supply chain needed for training.

OpenAI’s Nick Turley, head of the ChatGPT development team, described AI as an "existential threat" to news sites, noting that as AI products become more capable they may become direct substitutes for original content.

Legal context

The documents surface amid a lawsuit filed by The New York Times against Microsoft and OpenAI in late 2023, joined by 11 other publishing groups, accusing the firms of copyright infringement by harvesting and using millions of articles without permission. Both companies argue their use falls under "fair use" and claim the models do not directly replace original works.

The U.S. District Court for the Southern District of New York is reviewing motions for summary judgment while the related documents are being released incrementally.

These developments underscore the ongoing tension between the media sector and AI developers over content rights, prompting debate on how to balance innovation with protecting creators' interests.