Leaked Memos Reveal Microsoft Exec Called AI Training 'Largest Labor Theft in History'

Court filings submitted by The New York Times in its lawsuit against Microsoft and OpenAI revealed internal memos where a Microsoft executive characterized training AI on web content as potentially the largest-scale labor theft in human history. The documents also revealed OpenAI internal communications acknowledging that AI chatbots act as substitutes for content publishers, posing an existential threat to their business models. These internal acknowledgments undermine the public stance of AI developers who claim that training models on public web data constitutes 'fair use.' The evidence could heavily influence ongoing copyright litigation and accelerate regulations requiring tech companies to license copyrighted data for AI training. Internal Microsoft data disclosed that Copilot resulted in up to a 93% drop in click-through rates to The New York Times compared to traditional Bing search. Furthermore, messages among OpenAI staff showed awareness that users rarely click source links embedded in chatbot answers, and one engineer even shared a technique to bypass the newspaper's paywall.