AI Watchdog Datasets

AI Watchdog: VidGen-1M

Editor’s note: This search tool is part of The Atlantic’s investigation into how YouTube videos are taken to train AI tools. This work is part of AI Watchdog, The Atlantic’s ongoing investigation into the generative-AI industry.

VidGen-1M is a collection of 1 million 720p video clips paired with text captions. It was created by researchers at the Shanghai Academy of AI for Science and at Fudan University, in Shanghai, and released in 2024. To create the data set, the team pulled 3.8 million videos from Microsoft’s HD-VILA-100M data set and used an automatic process to generate captions. The data set is hosted on Hugging Face, an AI-development hub. This data set has been used by Amazon and Nvidia.