Newly unsealed court documents from the New York Times lawsuit against OpenAI and Microsoft reveal internal concerns regarding the impact of AI on the web. Internal records suggest that the companies acknowledged their data scraping practices could create a doom loop, potentially damaging the economic foundations of content creators and publishers.
Brent Hecht, Microsoft’s Director of Applied Science, described the harvesting of data as the largest theft of labor in human history. Microsoft stated that these comments represent an individual perspective and do not reflect the company’s official views. Jordan Usdan, GM for Data Strategy and Ops at Microsoft AI, characterized Hecht’s role as one focused on theoretical and futuristic viewpoints.
Internal documents from OpenAI indicate that GPT-4 was capable of memorizing and regurgitating copyrighted material verbatim. An OpenAI representative admitted to being unaware of efforts to remove paywalled content from training data. Furthermore, OpenAI’s Nick Turley noted that chatbots reduce the incentive for users to click on original source links.
Microsoft acknowledged that large language models act as a product that destroys its own supply chain by substituting for the labor that creates training data. OpenAI experts estimated that search referral traffic to sites like the New York Times could decline by as much as 60 percent due to AI-generated summaries.
