Unredacted NYT filings: a Microsoft director called AI scraping 'the largest theft of labor in human history'

TechCrunch·high signalXBlueskyLinkedInCopy link

Newly unredacted filings in the New York Times case against OpenAI and Microsoft show Microsoft applied science director Brent Hecht writing in a January 2023 internal memo that the practice was 'an astonishing theft of unprecedented proportions.' The filings put numbers on the training data: over 91,692 copies of works from NYT, Daily News and CIR in mid-training sets, more than 2 million nytimes.com documents from Common Crawl, and at least 160,903 unique publisher works in 'Project Mango.' Microsoft's own analysis found Copilot cut NYT click-through rates by as much as 93% versus traditional search.