TikTok Metadata for 5.6 Billion Videos Posted Free on Hugging Face

A developer posted metadata for about 5.6 billion public TikTok videos on Hugging Face, free under a non-commercial license, exposing captions, hashtags, shop IDs and engagement counts. Operators should audit exposure and review scraping rules.

What happened

A developer using the handle hashfunction has posted metadata for about 5.6 billion public TikTok videos on the open-source AI repository Hugging Face. The dataset appears under the datasocial account and is free to download under a non-commercial license. It runs from July 2014 through October 2026.

The material is metadata, not video footage. According to the dataset card, each row carries the caption, hashtags, sound ID, on-screen text, TikTok Shop product and seller IDs, and engagement counts for views, likes, comments, shares and saves. The developer's write-up says the information came from TikTok's private mobile API with no login.

TikTok's terms bar automated scraping without written approval. The same dataset card routes commercial use, creator profiles and daily updates to datasocial.ai, which also sells the scraper's source code for $1,699.

Key facts

• About 5.6 billion public TikTok videos have metadata posted on Hugging Face by developer hashfunction under the datasocial account.
• The dataset covers July 2014 through October 2026 and is free under a non-commercial license.
• Fields include caption, hashtags, sound ID, on-screen text, TikTok Shop product and seller IDs, and engagement counts for views, likes, comments, shares and saves.
• Data came from TikTok's private mobile API with no login, according to the developer's write-up; TikTok terms prohibit automated scraping without written approval.
• datasocial.ai handles commercial use, creator profiles and daily updates, and sells the scraper's source code for $1,699.

Our analysis

The release is significant because it converts public TikTok activity into a structured, queryable dataset at a scale of about 5.6 billion videos. The inclusion of TikTok Shop product and seller IDs alongside engagement counts suggests the data may be used for competitor research, influencer vetting and e-commerce intelligence, not just archiving.

The split between a free non-commercial license and paid offerings on datasocial.ai likely functions as a lead-generation funnel: the Hugging Face dataset builds visibility, while commercial access, creator profiles, daily updates and the scraper source code are monetized. This is different from many past scraping stories because the storefront is explicit about selling ongoing access and source code for $1,699.

Because the dataset is metadata rather than footage, copyright concerns may be lower, but platform terms still matter. TikTok prohibits automated scraping without written approval, so brands and creators that reuse this data for paid campaigns or client reports could face terms-of-service or contractual risk. The presence of TikTok Shop identifiers also suggests seller and affiliate activity is exposed beyond what a normal profile view would show.

What it means for operators

• Audit exposure: search the Hugging Face dataset for your brand, top creators and TikTok Shop accounts to see what metadata is publicly available and which fields are sensitive.
• Respect license boundaries: do not use the free non-commercial dataset in paid client work, ad targeting or commercial reports; if you need commercial use, negotiate with datasocial.ai or use TikTok-approved APIs.
• Review creator and affiliate contracts: add clauses on scraping, metadata reuse and competitive analysis, especially because TikTok Shop product and seller IDs are included.
• Monitor datasocial.ai updates: daily updates and creator profiles may reveal how competitors track your account performance, product links and engagement patterns.
• Treat scraped data as a compliance risk: buying or using scraped TikTok data from third parties could violate TikTok terms unless you verify the source and written approval.