Official Nvidia press releases and public statements

Nvidia Scrapes 80 Years of YouTube, Netflix Videos Daily for AI Training Despite Ethical Concerns

Aug 5, 2024, 05:17 PM

Leaked documents and internal communications reveal that Nvidia has been scraping massive amounts of video content from platforms like YouTube and Netflix to train an unreleased foundational AI model. The scale of the data collection is described as 'a human lifetime' or 80 years of videos per day. Despite legal and ethical concerns raised by staff, the company reportedly had clearance from the highest levels of management to proceed with the data scraping. The leaked information, reported by 404 Media, includes emails, Slack conversations, and other documents, highlighting the extensive use of copyrighted content and academic datasets for AI training.