A Massive YouTube Dataset to Train AI Systems: An Understudied Dataset with Hundreds of Million Youtube Subtitles and a Lookup Tool
More than 170,000 YouTube videos are part of a massive dataset that was used to train AI systems for some of the biggest technology companies, according to an investigation by Proof News and copublished with Wired. Apple, Anthropic, Nvidia, and Salesforce are among the tech firms that used the “YouTube Subtitles” data that was ripped from the video platform without permission. The training dataset is a collection of subtitles taken from YouTube videos belonging to more than 48,000 channels — it does not include imagery from the videos.
There are clips from ABC News, The New York Times and other news outlets in the dataset. There are a lot of videos from The vee in the dataset.
As part of its investigation, Proof News also released an interactive lookup tool. You can use its search feature to see if your content — or your favorite YouTuber’s — appears in the dataset.
She wasn’t going to say what the data was used for, but it was publicly available or licensed. When pressed about the matter by the Journal, she said she wasn’t sure.
Why AI companies should pay for their use? David Pakman, CEO of Proof News, YouTube Subtitles, and Founder of PewDiePie
“We have terms and conditions, and we would expect people to abide by those terms and conditions when you build a product, so that’s how I felt about it.”
Some of the videos found by Proof News were taken by MrBeast (289 million subscribers), Jacksepticeye (nearly 31 million subscribers, 377 videos taken), and PewDiePie (33 million subscribers, 337 videos taken). The material used to train Artificial Intelligence promoted certain conspiracy theories.
“No one came to me and said, ‘We would like to use this,’” said David Pakman, host of The David Pakman Show, a left-leaning politics channel with more than 2 million subscribers and more than 2 billion views. More than 160 of his videos have been uploaded to the YouTube Subtitles training dataset.
Four people work full-time on the enterprise, which posts multiple videos each day, as well as making and producing other content for other platforms. If AI companies are paid, Pakman said, he should be compensated for the use of his data. The agreements were penned to be paid for use of work by the media companies.
“This is my livelihood, and I put time, resources, money, and staff time into creating this content,” Pakman said. “There’s really no shortage of work.”
