Does TikTok keep Content Credentials on a downloaded AI video?

TikTok says it is "working to attach" Content Credentials to TikTok content, and that they "will remain on content when downloaded or even shared on other platforms", verifiable with the C2PA Verify tool (read 2026-10-03). That is a stated goal in progress, not a guarantee for a given file you download today.

The only reliable answer for your own clip is to download it and test it.

What has TikTok actually said?

On its transparency page, TikTok says it automatically labels AIGC made with TikTok AI effects, and labels AIGC from certain other platforms by reading C2PA Content Credentials. It says it is a C2PA General Member and joined the Content Authenticity Initiative. The attach-and-keep sentence is phrased as work in progress.

The May 9, 2024 newsroom post said TikTok had started reading Content Credentials on images and videos, with audio-only coming soon, and that "over the coming months" it would attach Content Credentials to TikTok content so they remain on download. The March 10, 2026 post speaks of invisible watermarking, tested because labels may be removed on re-upload or edit (read 2026-10-03).

Why test a file instead of trusting the page?

Between a 2024 promise of coming months and a present-tense page that still says working to attach, there is no dated statement we found that says rollout is complete for every video. So the safe assumption is that credentials may or may not be on the file you save from the app.

Testing is quick. Download your published clip, open it in C2PA's Verify tool named on TikTok's page, and note the result with the date. Our post on checking a delivered file gives a step-by-step.

Where does Sume come in?

Sume's docs describe job artifacts on media.sume.com, and do not say that Sume reads, writes or preserves Content Credentials. We cover that in what Sume adds to AI video. So if credentials appear on a TikTok download, they came from TikTok's pipeline, not from Sume.

That separation helps your records. Keep the Sume job id and artifact URL for what you uploaded, then separately log what the verify tool showed for the downloaded copy.

• Upload date, post URL and the label you set.
• Sume job id and artifact URL of the uploaded file.
• Verify tool result for the downloaded copy, with the date you tested.
• A note if credentials are absent: not proof of anything, only a fact about that file.

What if credentials are missing?

Their absence does not mean TikTok failed to label the video, since labeling can come from creator labels, detection or TikTok's tools, and TikTok says it shows which. Check the label in the app, not just the file.

If a clip needs durable provenance, burn a visible disclosure into the picture, which survives any metadata stripping; see our verification checklist for run output.

How do you write the test into your routine?

Make it a monthly check on one representative clip. Upload a short test video from your pipeline, download it from the app after it processes, run the verify tool, and save the result with the date. If the result changes between months, that is the useful signal: TikTok's pipeline moved.

Keep the test clip separate from campaign content so a failed or odd result never touches a live post. Note whether the test file had credentials before upload; Sume's docs make no claim that its outputs do, so the source file may carry none at all.

This gives you your own answer to the question the headline asks, which is worth more than a vendor sentence about intent.

Can you rely on detection of other platforms' files?

TikTok says it reads Content Credentials to label AIGC from certain other platforms. That is about files coming in. Sume's docs do not describe credentials on outputs, so a Sume file is unlikely to be recognized by that route. Do not expect an automatic TikTok label from a Sume file, and set the label yourself at upload.