Remove C2PA Content Credentials From a Video

Last updated November 12, 2026

C2PA Content Credentials in a video are stored in a metadata box inside the MP4 or MOV file, separate from the picture and sound. StripShot removes that box in your browser without re-encoding, so the picture and sound data are not changed. It does not remove watermarks in the content itself, such as Google's SynthID, and it does not change what a platform's own detector decides.

MP4, MOV and M4V, up to 500 MB. No re-encoding. The file is not uploaded. Scanning a video is free. Open the full-screen video cleaner.

Where Content Credentials sit in an MP4 or MOV

An MP4 or MOV file is a list of boxes. One box holds the picture and sound. Others describe the file: its length, its tracks, the app that wrote it. The C2PA specification puts the Content Credential in a box of its own, a uuid box with the ID d8fec3d6-1b0e-483c-9297-5828877ec481, held by the file itself and not by a track (C2PA 2.2, Appendix A.5.1).

Because the credential is a separate box, it can be taken out without touching a frame. That is the whole method: StripShot reads the list of boxes, leaves out the metadata ones, corrects the offsets that point at the picture and sound, and writes the file back.

What comes out with it

A video that carries Content Credentials usually carries other metadata too. A clean removes all of it in one pass:

• The C2PA box, at the top of the file and inside the movie box or a track description.
• XMP blocks, which can hold the tool name and the IPTC digital source type.
• The user data and metadata boxes: software and tool names, titles, comments, location, device make and model.
• Creation and modification dates, which are reset.
• Tracks that are not picture or sound: timed metadata, subtitles, chapters and timecode.

The scan shows what it found before anything is removed, and scanning is free. For every AI signal a video file can carry, not only C2PA, see the guide to removing AI metadata from video.

What stays

• Watermarks in the content. Google says SynthID "embeds digital watermarks directly into AI-generated images, audio, text or video" and is designed to hold up through cropping, filters, frame rate changes and lossy compression (Google DeepMind). No metadata tool removes it.
• Visible marks: a logo or a caption in the picture.
• A platform's own detection. TikTok says it labels with "our own detection models" as well (TikTok, Nov 19, 2025), and YouTube lists "content that our internal systems detect" (YouTube Help). Neither company has published how that detection works. A clean changes metadata only, not the picture or sound.
• Your duty to disclose. Where a platform requires you to label realistic AI video, that rule applies whatever the file carries.

Which platforms say they read C2PA in video

• TikTok. It reads Content Credentials to recognize and label AI-generated content. TikTok said the capability "started rolling out today on images and videos" on May 9, 2024 (TikTok). More on TikTok's AI label on videos.
• YouTube. Its help page says YouTube may automatically apply an AI label for "content that contains C2PA metadata" (YouTube Help).
• LinkedIn. "Image and video content that is cryptographically signed using C2PA Content Credentials will be noted with the C2PA icon" (LinkedIn Help).

The platform tracker keeps the current rule for each platform, with dates.

Why this is a before-upload question on YouTube

YouTube lets creators change an AI disclosure in most cases, but not all. Its help page lists content containing C2PA metadata among the cases that can not be adjusted (YouTube Help), and its May 27, 2026 post says the disclosure stays permanent for content containing C2PA metadata that indicates it was fully generative AI (YouTube). So a credential in the file at upload is one of the cases where YouTube says the label can not be changed later. The YouTube label guide has the details.

Does ffmpeg remove C2PA from a video?

In our test it did. On November 12, 2026 we ran ffmpeg 9.0.2 with -c copy -map_metadata -1 on a test MP4 that had a C2PA box at the top of the file. The output had no C2PA box and no XMP. It did carry ffmpeg's own encoder tag, because ffmpeg writes a new container. That was one test file built for the purpose, not a signed export from a generator, so check your own result.

The difference is in what you can see. StripShot scans first, lists what the file carries, and then removes it in the browser with nothing to install. If you already use ffmpeg, scan its output here for free to confirm what is left.

Which AI video tools write C2PA?

Vendors change this and rarely document it, and we have not opened a vendor page that settles it for video. The reliable check is the file itself: drop your export into the scanner above. It reports Content Credentials when the box is there, and whether the credential says the video was made with AI.

Formats and limits

• MP4, MOV and M4V files up to 500 MB.
• Fragmented MP4 is not supported. Export a standard MP4 from your tool first.
• WebM is not supported.
• The free plan cleans one video every 30 days. Pro ($9/month) cleans 10 a day and Full Access ($19/month) has no daily limit.

Check the result

After you save the clean copy, drop it back into the scanner. A clean file reports no Content Credentials, no XMP and no tool names. For a photo, use the free checker.

Related guides