Court filings reveal internal Microsoft and OpenAI concerns over AI data scraping impact
Just the facts

Court filings reveal internal Microsoft and OpenAI concerns over AI data scraping impact

Summary

Unsealed documents in the New York Times lawsuit show Microsoft and OpenAI executives discussed the potential negative effects of large-language-model training on web publishers and the broader content ecosystem.

New York court filings released in the New York Times case against OpenAI and Microsoft contain internal memos that acknowledge the companies’ large-scale web scraping could undermine the economics of online publishing. A Microsoft internal document described the AI content strategy as creating a “doom loop” that might hurt both model performance and the web’s content supply chain. The memo said the situation was “highly unusual” because the end product threatens the economic foundations of its essential suppliers.

Microsoft’s Director of Applied Science, Brent Hecht, is quoted in the filings as calling the data harvesting for ChatGPT and Copilot “the largest theft of labor in human history” and criticizing the company’s fair-use defense. Microsoft spokesperson Alex Haurek said the comments reflect an individual employee’s perspective and do not represent the company’s official view. Jordan Usdan, a Microsoft AI executive, described Hecht’s remarks as “divergent, academic, and forward-looking” and not representative of Microsoft policy.

Internal OpenAI communications also discuss the risk of the models reproducing copyrighted material verbatim. Employees noted that preventing memorization was important to reduce potential copyright violations, while acknowledging that GPT-4 had “memorized a ton of data.” The filings cite examples where ChatGPT generated text that closely matched articles from several news outlets.

Both companies have publicly stated that pay-walled content should be licensed, but an OpenAI representative admitted uncertainty about any systematic effort to filter such material from training data. OpenAI policy director Jack Clark described the technology as “substituting for the labor of the people that define the culture of society,” and internal notes referred to ChatGPT as “the modern newsstand."

Microsoft’s legal team emphasized that executive statements about changes in how people find information should not be taken as conclusions on copyright issues, which the company addresses separately in its filings.

Keep reading — for free

Create a free account to follow the news. No card required.