Microsoft Told Employees to Label OpenAI Scraping Project 'Confidential'

A court document unsealed on October 16 shows that Microsoft Corporation had ordered employees to keep its website scraping project confidential; the project collected artificial intelligence training data for OpenAI. In this project, codenamed "Project Mango," OpenAI paid Microsoft to develop and operate a web crawler used to copy webpage content; this content was then used to train models including GPT-4o. News publishers stated this in a partially unsealed motion for summary judgment.