Unsealed legal documents from a motion for summary judgment filed by news organizations, led by The New York Times, reveal internal communications from Microsoft and OpenAI regarding the use of scraped news content for training large language models (LLMs). Internal documents show Microsoft Director of Applied Science Brent Hecht characterized news scraping as "an astonishing theft of unprecedented proportions" and the "largest theft of labor in human history," while noting that the practice makes a "complete mockery of the idea of ‘fair use.’" OpenAI executives similarly expressed concerns, with ChatGPT head Nick Turley describing chatbots as "largely substitutive" and predicting an existential threat to publishers. This substitution is reflected in significant traffic declines, with some news plaintiffs reporting click-through rate drops between 83% and 94%. Furthermore, the filings allege that OpenAI obtained a dataset of 1.8 million New York Times articles through a third party despite knowing it was restricted to non-commercial use. Plaintiffs argue that the companies' products provide verbatim article excerpts through prompts such as "what's the next line?" or requests for bias ratings, creating a "doom loop" where AI models cannibalize the "content supply chain" they depend upon for reliable training data.