Full Breakdown
Internal Documents Show Microsoft and OpenAI Acknowledge “Doom Loop” Threat to News Industry
By Drooid · · How we work
Background of the New York Times Lawsuit
The New York Times filed a copyright infringement lawsuit against OpenAI and Microsoft in late 2023, alleging that the companies trained large language models (LLMs) on millions of the newspaper’s articles without permission or compensation. The case has remained active for nearly three years. In a recent motion for summary judgment, the publishers presented unsealed internal memos and deposition excerpts from both firms.
Key Internal Assertions
Multiple internal documents reveal that senior executives described the AI training approach as a direct threat to the news ecosystem:
- OpenAI’s head of ChatGPT Nick Turley labeled the technology an “existential threat” to publishers and said the products were “largely substitutive, period.”
These statements are presented by the plaintiffs to argue that the companies knew their models would compete with the very content they harvested.
Data on Publisher Traffic Decline
The filing cites internal Microsoft analytics showing dramatic drops in click-through rates when news articles appear in AI-generated answers:
- For the Times, Bing-Chat click-through rates were 83 % to 93 % lower than traditional Bing searches.
- Other plaintiffs reported declines ranging from 51 % to 94 %.
OpenAI’s training datasets reportedly contain more than 91,692 copies of works from the Times, Daily News, and the Center for Investigative Reporting, while a Common Crawl-derived set includes over 2 million documents from nytimes.com alone.
Official Statements from Microsoft and OpenAI
- He added that Microsoft’s formal court filings argue the AI uses are “transformative” and “consistent with copyright law.”
- OpenAI has not provided a comment to TheWrap, but its legal team maintains that the training process falls under the fair-use doctrine, arguing that the models are “transformative” and that the data are used for “research and scholarship.”
Both companies emphasize that the quoted internal language does not constitute official policy, even as the court documents show the language was part of internal strategy discussions.
Verbatim Quotes
- “It is highly unusual that an end-product threatens the economic foundations of its essential suppliers, but that is the situation we have created for our LLM business with respect to its ‘content supply chain.'” — Frankly
- “Millions of people around the world will soon consider large models ‘hoovering up’ all their work to be an astonishing theft of unprecedented proportions” — Applied Science Brent Hecht, microsoft executive
What’s Next
The court will consider the summary-judgment motion in the coming weeks. The judges must decide whether the internal admissions undermine the defendants’ fair-use defense and whether the alleged “doom loop” constitutes a market-harming effect sufficient to deny copyright protection for the training practices. The outcome could shape how AI developers source and compensate copyrighted material in the future.