Recent findings by Proof News reveal that leading tech companies such as Apple, Nvidia, and Anthropic have used YouTube content to train their AI models without informing or compensating the creators. This revelation has caused significant outrage among YouTube content creators who discovered their work was being utilized without their consent.
The investigation uncovered that subtitles from 173,536 YouTube videos across more than 48,000 channels were included in a dataset named YouTube Subtitles. This dataset, used by major tech firms to train AI, includes transcripts from channels like Khan Academy, MIT, and prominent media outlets such as NPR and the BBC.
Popular YouTubers like MrBeast, Marques Brownlee, Jacksepticeye, and PewDiePie had their content used without permission. David Pakman, whose channel The David Pakman Show had nearly 160 videos included, expressed his frustration, saying, “No one came to me and said, ‘We would like to use this.’ This is my livelihood, and I put time, resources, money, and staff time into creating this content.”
Dave Wiskus, CEO of the creator-owned streaming service Nebula, condemned the practice, calling it “theft” and “disrespectful.”
EleutherAI, the creators of the dataset, did not respond to requests for comment. The dataset consists of plain text subtitles and translations from YouTube videos, in apparent violation of YouTube’s terms of service. Despite these terms, tech giants like Apple, Nvidia, and Salesforce have admitted to using the dataset as part of a larger compilation called the Pile.
Jennifer Martinez, spokesperson for Anthropic, confirmed their use of the dataset, emphasizing that YouTube’s terms cover direct platform use, which is distinct from using the Pile dataset.
This situation underscores the need for clear regulations and ethical guidelines to protect content creators in the digital age. As AI technology continues to advance, ensuring that creators are fairly compensated and their intellectual property rights are respected remains crucial.