YouTube AI label and C2PA: check your video before you upload.

When YouTube shows an AI label

YouTube asks creators to disclose AI use in YouTube Studio: in the Attributes section, under "AI use", you select Yes if the video meets the disclosure requirements (YouTube Help). The requirement covers using AI "to meaningfully alter or generate photorealistic content", including music that is the main focus of a video.

YouTube may also apply an AI label automatically, for three kinds of content (YouTube Help):

• Content made with YouTube's own generative AI tools
• Content that contains C2PA metadata
• Content that YouTube's internal systems detect as AI-generated or altered

The detail that matters most: YouTube says labels on "content containing C2PA metadata" cannot be adjusted by the creator.

Why check the file first

Because a label triggered by C2PA can't be changed after upload, the time to look is before. An exported video can carry a C2PA manifest from an AI video tool, from an editing app that used a generative feature, or from a source clip you dropped into the timeline. This tool shows you whether the file has one.

1. Drop the MP4 or MOV you plan to upload.
2. Check the report for a C2PA line. In video files the manifest sits in a top-level uuid box, often with an XMP packet that refers to it.
3. If the video is not AI content and you don't want the manifest published, download the cleaned copy. The video and audio streams are copied without re-encoding, so YouTube receives the same picture and sound.

For editors: check your source clips

A manifest in one source clip does not automatically end up in your final export. Whether it does depends on the editing app. That is why the file worth checking is the one you actually upload, not the clips that went into it. If your final export carries a C2PA manifest you didn't expect, look at which tool names the report mentions. It usually points to the app or AI feature responsible.

Large exports are fine. The tool reads only the small index of a video file and never loads the video data into memory, so multi-gigabyte uploads are checked in moments and cleaned without a size limit.

Disclosure is not a penalty

YouTube states that "disclosing AI content won't limit a video's audience or impact its eligibility to earn money" (YouTube Help). It also warns that creators who consistently choose not to disclose may get a label applied manually, or face penalties including removal of content or suspension from the YouTube Partner Program.

For realistic AI-generated videos, the safe and honest route is simply to disclose. Metadata removal makes sense when a manifest is on the file for the wrong reason, or when you want to strip location, device and app details before publishing.

The other C2PA label: "Captured with a camera"

C2PA can also work in your favour on YouTube. To get the "captured with a camera" label, creators must record with tools that have built-in C2PA support (version 2.1 or higher), and the content must not have edits to sound or visuals (YouTube Help). If you shoot with such a camera and want that label, upload the original and do not clean it. YouTube adds that a missing label "doesn't mean the content has modified audio or visuals".

What cleaning does not remove

• Invisible watermarks in the frames, such as SynthID in video made with Google's Veo. See SynthID explained.
• Labels already applied to a published video.
• Whatever YouTube's internal systems detect from the content itself.

For the technical side of MP4 and MOV metadata, see the video metadata remover.