Unsealed court documents in the New York Times OpenAI lawsuit reveal that both tech giants internally acknowledged their artificial intelligence models were creating a "doom loop" for the web. Internal communications from a 92-page filing show executives warning that scraping publisher data constituted a massive theft of labor, even as they pushed forward to commercialize the technology. This revelation is critical for digital publishers, content creators, and legal professionals tracking the boundaries of fair use, as it provides concrete evidence that the companies knew their products would devastate the economic foundations of the internet.
The most striking admissions come from Brent Hecht, Microsoft's Director of Applied Science. Hecht characterized the data harvesting powering ChatGPT and Copilot as the "largest theft of labor in human history," noting that the company's legal defense made a "complete mockery of the idea of fair use." Microsoft has since attempted to distance itself from these statements. Jordan Usdan, GM for Data Strategy and Ops at Microsoft AI, stated in a separate filing that Hecht holds "divergent, academic, and forward-looking views" that do not represent the company's official stance.
Satya’s testimony and Microsoft’s position in this case are perfectly consistent. He spoke to broad principles and changes underway in how people find and consume information.
- Alex Haurek, Microsoft
Despite the corporate distancing, the internal documents highlight a widespread awareness of the damage being done to the content supply chain. An internal Microsoft document explicitly warned that their AI content strategy had started a "doom loop" that would ultimately hurt both the performance of their Large Language Models (LLMs) and the entire web. Meanwhile, OpenAI cofounder Greg Brockman was noted as being focused on the "gazillions" of dollars the company could potentially make through commercial AI, contrasting sharply with the company's original altruistic mission.
Internal Admissions and Publisher Impact
The 92-page filing details several specific areas where OpenAI and Microsoft executives, including Sam Altman and Satya Nadella, recognized the destructive nature of their products:
- Copyright Regurgitation: Employees admitted that GPT-4 "memorized a ton of data and therefore will be insanely good at regurgitation." The filing cites examples of ChatGPT outputting verbatim text from the New York Times, Mercury News, The Denver Post, LifeHacker, and Eurogamer.
- Paywall Bypassing: While Nadella stated that "anything that is paywalled should be licensed," an OpenAI representative admitted to being unaware of any effort to remove paywalled content from training data.
- Traffic Destruction: OpenAI Policy Director Jack Clark warned they were creating systems that substitute for human labor. Furthermore, OpenAI’s own media experts attributed a drop in referral traffic directly to AI summaries, estimating that search referrals could fall by as much as 60 percent.
The Legal Vulnerability of Market Substitution
These unsealed documents fundamentally alter the trajectory of the ongoing copyright battles by attacking the core pillar of the "fair use" defense. A primary factor in determining fair use is whether the new product serves as a market substitute for the original work. By internally describing ChatGPT as "the modern newsstand" and OpenAI's Nick Turley admitting that users have "no good reason to click" on source links, the companies have essentially documented their own market substitution.
If courts rule that these internal warnings invalidate the fair use argument, the tech industry will be forced into mandatory licensing agreements, permanently altering the economics of AI development. The admission that LLMs are a product that "destroys its own supply chain" proves that the current scraping model is not just legally precarious, but structurally unsustainable for the future of the web.
