OpenAI and Microsoft Knew AI Was Starting a Web Doom Loop

Newly unsealed court filings in the New York Times' ongoing copyright lawsuit against OpenAI and Microsoft show that employees at both tech giants harbored deep internal concerns about the impact of generative AI. According to the documents, Microsoft's own internal analysis warned that its artificial intelligence content strategy had initiated a "doom loop" that would simultaneously degrade the performance of its large language models and damage the broader internet. The files reveal a shared internal recognition that these systems act as a direct substitute for the very content creators that supply their training data.

Among the most striking disclosures are comments from Brent Hecht, Microsoft's director of applied science, who characterized the wholesale scraping of web data as the "largest theft of labor in human history" and argued that the company's legal defense made a "complete mockery" of fair use. Microsoft has sought to distance itself from Hecht, with general manager Jordan Usdan describing his views as divergent and academic. However, other internal communications show OpenAI employees admitting that GPT-4 had "memorized a ton of data" and was highly prone to verbatim regurgitation, despite public claims of preventing copyright violations.

The documents also highlight how the companies anticipated the economic fallout for publishers. OpenAI's head of ChatGPT, Nick Turley, noted that once users receive a direct answer from a chatbot, they have "no good reason to click" on original source links. OpenAI's own media and economic experts estimated that AI-generated summaries could slash search referral traffic to publishers by as much as 60 percent. Meanwhile, Microsoft documents openly acknowledged that large language models represent "a product that destroys its own supply chain" by starving the creators who produce the original data.

Despite Microsoft CEO Satya Nadella publicly stating that paywalled content should be licensed, an OpenAI representative admitted in the filings to being unaware of any internal efforts to filter out paywalled material. Instead, the documents depict an aggressive push for commercialization, with OpenAI co-founder Greg Brockman focused on the potential to make "gazillions" of dollars. Representatives for Microsoft have since downplayed the unsealed comments, stating they represent individual perspectives rather than official legal positions.