Newly unsealed court documents from an ongoing copyright lawsuit reveal that a Microsoft executive privately described AI training practices as the “largest theft of labor in human history,” according to filings from The New York Times. The three-year-old legal battle, which also targets OpenAI, centers on whether artificial intelligence firms can legally scrape copyrighted material to train large language models without permission or payment.
Internal Admissions and Fair Use Challenges
The unredacted material details how technology companies allegedly bypassed paywalls, built mass training datasets, and stripped copyright notices from material to build generative AI products.
Internal documents show that OpenAI leadership recognized products like ChatGPT posed an “existential threat” to publishers and journalists, noting the chatbots are largely substitutive for original reporting. OpenAI President Greg Brockman described the models as excellent at news, while Microsoft CEO Satya Nadella testified in a deposition earlier this year that conversing with chatbots substitutes for visiting an underlying news website.
Meanwhile, OpenAI’s head of ChatGPT, Nick Turley, acknowledged internally that AI products would become increasingly substitutive over time.
The Economic Doom Loop for Publishers
The filings highlight a direct economic conflict between AI platforms and their content suppliers. An internal Microsoft presentation written in January 2024 by Brent Hecht, Microsoft’s director of Applied Science, described a “doom loop” where Microsoft’s Copilot answer engine caused click-through rates for The New York Times’ domain to drop by up to 93% compared to traditional Bing search.
Hecht wrote that it is highly unusual for an end-product to threaten the economic foundations of its essential suppliers. Additional Microsoft documents cited in the filings warn of a real risk that generative AI could disrupt the employment of the workers who generated the training data.
News groups filing the motion argued that AI companies are trapped in a prisoners’ dilemma, where individual firms benefit from taking free content while the industry as a whole suffers from the degradation of its content supply chain. Brockman was motivated by the substantial commercial potential of the technology, according to the legal filings.
Legal Arguments and Industry Practices
The lawsuit details aggressive data acquisition methods.
Microsoft also faced scrutiny for allegedly violating industry norms by selling a dataset purchased for Bing search to OpenAI as training data without consulting news publishers. The legal challenge follows a recent brief filed by the Trump administration in defense of OpenAI’s unlicensed use of copyrighted material.
Publishers argue that a clear judicial ruling against unlicensed scraping is necessary to establish market parity. Without requirements to license content, both news organizations and AI firms face severe financial risks as automated summaries replace web traffic.