Newly unsealed court documents revealed internal discussions at OpenAI and Microsoft about the potential impact of artificial intelligence on news publishers, including concerns that AI-powered products could reduce traffic to news websites and compete with traditional journalism.
The documents form part of the copyright dispute between The New York Times and the two technology companies. Publishers argue that their journalism was used to train AI systems without permission and that those systems could ultimately undermine the businesses that produced the underlying content. OpenAI and Microsoft have disputed those arguments and maintain that AI training can qualify as fair use.
According to the unsealed filings, Nick Turley, who led the ChatGPT team, described AI as an “existential threat” to publishers in a June 2023 internal communication. In a February 2024 message cited in the court documents, Turley reportedly said OpenAI's products were becoming increasingly capable of replacing some traditional ways of accessing information. The filings also cite an OpenAI engineer questioning whether users would still click through to original articles even when links were prominently displayed.
The publishers argue that such comments support their position that AI products can compete directly with news organizations rather than simply offer a different way to access information.
The filings also contain internal comments from Microsoft. Brent Hecht, the company's director of applied science, warned about a potential ‘doom loop’ in which AI-generated answers reduce visits to websites that produce original content.
The concern was that lower web traffic could weaken publishers' advertising and subscription businesses. Over time, that could reduce the amount of original material available online, potentially affecting the broader content ecosystem AI systems rely on.
Microsoft CEO Satya Nadella also testified that chatbot conversations had substituted for visits to some publisher websites by providing information directly through AI platforms.
Also Read: OpenAI AI Agents Probed Hugging Face Before July Hack
The newly revealed communications are being used by publishers to challenge OpenAI and Microsoft's fair-use arguments. The publishers allege that the companies copied large volumes of copyrighted journalism to train AI systems and that the resulting products can compete with the original sources.
The filings also describe an exchange involving an OpenAI employee and company president Greg Brockman concerning a possible way to access New York Times content behind a paywall. The publishers cite the exchange as part of their broader claims over how copyrighted material was obtained.
However, the newly unsealed evidence does not itself establish that either company violated copyright law. The central legal question, whether the use of publishers' material for AI training qualifies as fair use remains before the court. OpenAI has continued to argue that its use of copyrighted material to develop AI systems is legally protected.