Skip to content
SnapFrom

TikTok's Terms Grant You a Download Licence and Ban the Tools That Use It

· · updated · 5 min read

TikTok’s Terms of Service contain two clauses that point in opposite directions. Both are current as of the 15 July 2026 revision, and reading them together explains more about this topic than any set of instructions.

Clause one: you have a licence to download

Section 3.5 says that depending on a creator’s sharing activities and settings, you grant every other user of the platform “a non-exclusive, royalty-free, perpetual and irrevocable, worldwide license to use Your Content, including to access, reproduce (e.g. to copy), distribute, share, download, adapt or make derivative works”[1].

Read that from the other side. When you watch someone’s public TikTok, their content came to you with a perpetual, irrevocable licence to download and adapt it, granted by them, through TikTok, to you. That is not a loophole or a reading - it is what the contract says, and the word “download” is in it explicitly.

Clause two: you may not use a tool to do it

Section 3.4, in the list of things you must not do, bars you from[1]:

scrape, crawl, export or otherwise extract any data or content in any form, for any purpose, from the Platform using any automated system or software, including automated “bots,” except as approved in writing

“In any form, for any purpose” is about as wide as that sentence can be drawn. Every third-party TikTok downloader is automated software extracting content from the platform. All of them are inside this clause.

So the licence in 3.5 is real, and essentially every practical means of exercising it is prohibited by 3.4. The only route TikTok leaves open is its own in-app save button, which is the one that burns the watermark in.

What this means in practice

Three things follow, and they are more useful than a set of steps:

  • Downloading a public TikTok is not, by itself, a copyright problem. The creator granted the licence when they posted publicly. That is genuinely different from the YouTube situation, where no such grant exists.
  • Getting the file with a tool is a terms problem, between you and TikTok, not between you and the creator.
  • Reposting is a different question again. Section 3.4 separately forbids using another user’s content “for commercial purposes unless permitted by TikTok USDS Joint Venture or the user”[1]. The download licence does not carry a commercial-use right with it.

A note on who “TikTok” is, because it changed and the terms reflect it: the US platform is now operated by TikTok USDS Joint Venture LLC, and the terms state they “completely replace any prior terms of service between you and TikTok Inc.”[1]. If you last read these terms under the old entity, you read a different contract.

Why the tools keep breaking

This is the part worth knowing, and the part no downloader’s own page will publish.

TikTok serves video from CDN URLs that are tied to a session and expire. Handing a downloader the raw media URL therefore does not work for long, which is why every functioning tool works from the post URL and re-resolves the stream each time.

Re-resolving means making a request that has to look like a browser. Not a user-agent string - the fingerprint of the connection. yt-dlp, the reference implementation, ships an --impersonate option built on curl_cffi that reproduces a named browser’s TLS and HTTP/2 characteristics, plus --cookies-from-browser to reuse a real logged-in session[2]. Those exist because header-level disguises stopped being sufficient.

Two failure modes follow, and between them they account for nearly every “it stopped working” report:

  • The impersonation target goes stale. It reproduces a specific browser version. When real Chrome moves on and the target does not, every user of that tool breaks simultaneously, which is why these outages look like the service died rather than degraded.
  • Session reuse does not survive scale. One person’s cookies work fine. The same cookies driving thousands of strangers’ downloads from a few server addresses look like exactly what they are.

The yt-dlp README notes that impersonating requests “may have a detrimental impact on download speed and stability”[2]. The reference implementation treats this as an ongoing cost, not a fixed problem, and anything built on top inherits that.

The watermark, honestly

The in-app save writes TikTok’s moving watermark and the creator’s username into the video. That is the cost of the only route TikTok sanctions, and it is a deliberate design choice rather than an oversight: the watermark is how a clip stays attributable after it leaves the app.

If your reason for wanting a clean copy is to repost the video as your own, the watermark is working correctly and the terms above are against you. If your reason is editing footage you have rights to, the answer is to get it from the creator, who has the original.

Frequently asked questions

Is downloading a public TikTok illegal?

Copyright and terms of service are different questions. The licence in section 3.5 addresses the copyright side for public content[1]. The automated-extraction ban in 3.4 addresses your agreement with the platform[1]. You can be clear on one and not the other.

Can I download private or friends-only videos?

No, and no tool can. The licence in 3.5 is explicitly conditioned on the creator’s sharing settings[1].

Does downloading notify the creator?

No. Creator analytics report views and shares, not downloads by third-party tools.

Why do these services work one week and fail the next?

Almost always the impersonation target going stale or the service’s addresses being throttled, as described above. It is rarely anything you did, and reinstalling or switching browsers will not help.

Sources

  1. TikTok Terms of Service (United States), last updated 15 July 2026 - TikTok USDS Joint Venture LLC, accessed

    Supports: Section 3.5 grants each user a perpetual, irrevocable, worldwide licence to download other users' content subject to sharing settings, while section 3.4 forbids scraping, crawling, exporting or extracting any data or content by automated system or software without written approval, and forbids commercial use of another user's content without permission.

  2. yt-dlp README - yt-dlp project, accessed

    Supports: yt-dlp provides --impersonate, built on curl_cffi, to reproduce a specific browser's fingerprint, and --cookies-from-browser to reuse a logged-in session - and notes that impersonating requests can degrade download speed and stability.

Get new articles by email

One email per new article, at most twice a week. No other mail, and every message carries a working unsubscribe link. You will get a confirmation email first - we do not add an address that has not clicked it.