150,000+ YouTube Videos Were Used by Silicon Valley Giants to Train AI

Proof News, in collaboration with Wired, has published an extensive investigative report stating that numerous tech companies, including NVIDIA, Apple, Salesforce, and Anthropic, have used content from thousands of YouTube videos to train their AI models, completely ignoring YouTube’s rules against harvesting material from the platform without permission.

According to the investigation, Silicon Valley giants employed a service called YouTube Subtitles to access subtitles from 173,536 YouTube videos, sourced from over 48,000 channels, including Khan Academy, MIT, Harvard, The Wall Street Journal, BBC, late-night shows, and popular YouTubers like MrBeast, Marques Brownlee, Jacksepticeye, and PewDiePie.

The subtitles were then utilized as training data for the companies’ generative AIs, showing once again that when it comes to artificial intelligence, multi-billion companies are perfectly content with using tactics of questionable legality to gain an edge over their competitors in the AI race.