Public statements or legal filings by the YouTube creators

Apple, Nvidia, Anthropic, and Salesforce Used 170,000 YouTube Videos to Train AI Without Consent

Jul 16, 2024, 03:19 PM

A recent investigation by Proof News has revealed that major tech companies, including Apple, Nvidia, Anthropic, and Salesforce, have used subtitles from over 170,000 YouTube videos, siphoned from more than 48,000 channels, to train their AI models without the consent of the content creators. The dataset, curated by EleutherAI and known as 'the Pile,' includes videos from popular YouTubers such as MKBHD, MrBeast, John Oliver, and educational channels like Khan Academy. This practice has sparked a debate over ethical data-sourcing and copyright issues, as creators claim their videos were used without their knowledge. The investigation highlights that the dataset contains transcripts from prominent sources like WSJ, NPR, and BBC. Apple, which has previously emphasized paying for data used in AI training, is now facing scrutiny over these revelations.