A developer operating under the handle hashfunction has uploaded metadata for roughly 5.6 billion public TikTok videos to the open-source AI repository Hugging Face, where it is free to download under a non-commercial license. The dataset, published under the datasocial account, spans July 2014 through October 2026.

The release contains metadata rather than video footage. Each row includes the video's caption, hashtags, sound ID, on-screen text, TikTok Shop product and seller IDs, and counts for views, likes, comments, shares, saves, and downloads. The full archive weighs 460 GB, stored as Parquet files organized one per month. That footprint is modest for a corpus of this scale — a function of carrying metadata rather than video — and the one-file-per-month layout lends itself to time-based analysis.

Some fields reflect TikTok's own labels, such as whether a video is flagged as AI-generated and whether the platform keeps it off the For You page. The AI flag is frequently empty for older entries, which come from an archive — a gap that limits any analysis relying on that label for earlier years.

The dataset is not gated, meaning anyone can pull it, with no application, approval, or institutional affiliation required. Hugging Face's counter showed 1,181 downloads so far. It is also available online via datasocial.ai.

The release matters because AI developers want this kind of data, and TikTok has made it hard to obtain. Captions, hashtags, sound IDs, and engagement counts across billions of posts are raw material for models that predict what goes viral or what sells on TikTok Shop, which words and phrases click and which do not, and for language models learning how people write in short-form video.

TikTok's official route to such data is far narrower. Its Research Tools open to qualifying researchers at academic institutions in regions including the United States, the EEA, the UK, Canada, and Switzerland, plus some not-for-profit bodies in the EU, and require an application and approval — eligibility criteria that leave commercial AI developers without an official path to comparable data.

Broad data scraping is not permitted by the platform. Section 3.4 of TikTok's U.S. terms of service extracting data from the platform with automated software unless TikTok approves it in writing.

According to DataSocial's write-up, the collection reached TikTok's private mobile API using generated device identities that passed as Android phones, reverse-engineered request signatures, and a spoofed TLS handshake. The write-up says the system collected 3.23 billion creator profiles, 5.94 billion videos, and 2.8 billion comments in three weeks, and that no login or account was involved.

The free tier doubles as a storefront. The license is CC BY-NC 4.0, which permits sharing and adapting the data with attribution so long as the use stays non-commercial, and the dataset card describes it as free for research and personal use. A license of that kind governs how the files may be reused; it sits apart from TikTok's terms of service, which govern how the data was collected. Commercial use, creator profiles, and daily updates are routed to datasocial.ai, which also sells the scraper's source code for $1,699.

Scraping disputes are already playing out in court. Reddit sued Perplexity and three data-scraping firms in October 2025, alleging an "industrial-scale" scheme to harvest its content for AI training, as Decrypt reported. In July, a federal judge largely declined to dismiss the case, per Law360. The surviving claims include DMCA allegations that SerpApi circumvented Google's anti-bot protections. TikTok is not a party, and that case involves scraping through Google search results rather than a private mobile API, meaning its outcome will not directly settle the legal status of this collection.

As for the dataset itself, rows include captions, music used, and various label fields. There is no creator handle column, though the data carries captions, hashtags, and people mentioned in videos. The user ID, however, is available through Datasocial.

The data trove already has company. Copies of a separate 4.5-billion-video set, also described as collected from TikTok's mobile API, are on Hugging Face as well, free for non-commercial use. The code that collected it costs $1,699.