In a significant move for AI developers and researchers, a developer known as hashfunction has made a vast dataset of TikTok video metadata publicly available. According to the official information, this release, hosted on the open-source AI platform Hugging Face, offers a treasure trove of information for those looking to delve into TikTok's content landscape.
Overview of the Dataset
The dataset, published under the account datasocial, encompasses metadata for approximately 56 billion public TikTok videos, covering a timeline from July 2014 to October 2026. It includes essential details such as video captions, hashtags, sound IDs, and engagement metrics, providing a comprehensive overview of user interactions on the platform.
Significance of the Release
This release is particularly noteworthy as TikTok has increasingly restricted access to its data through official channels, making it difficult for developers to obtain similar information. By offering this dataset for free download, hashfunction has created a valuable resource for AI developers aiming to analyze trends and behaviors within the TikTok ecosystem.
On September 15, 2026, bdautomated released a dataset addressing AI agent statistics, providing clarity on AI adoption rates. This contrasts with the recent release of TikTok video metadata by hashfunction, highlighting the growing importance of accurate data in the AI landscape. For more details, see AI statistics.













