Microsoft director calls AI training 'the largest labor theft in human history'

Beating AI News Flash: A batch of internal materials has been unsealed in The New York Times' copyright lawsuit against OpenAI and Microsoft. Brent Hecht, Microsoft's Director of Applied Sciences, once called large-scale scraping of web content to train AI "an unprecedented act of stunning theft," and possibly "the largest labor theft in human history."

Nick Turley, then head of ChatGPT, also said that AI products would pose an "existential threat" to publishers, essentially serving as substitutes for content, and that the stronger the capabilities, the greater the degree of substitution. Microsoft CEO Satya Nadella acknowledged in testimony that chatbots would allow users to get answers directly, reducing visits to original sources.

Another Microsoft internal document called this model a "death loop": AI takes away content producers' revenue, high-quality content decreases, and ultimately this also affects AI's own training. The plaintiffs also cited Microsoft internal data stating that compared with traditional Bing search, Copilot brought 83% to 93% fewer clicks to news websites such as The New York Times.

OpenAI and Microsoft currently still argue that using copyrighted content to train AI constitutes fair use. Microsoft said Hecht's statement was only his personal opinion, did not represent the company's position, and was not a legal conclusion on copyright issues.