TikTok Automatically Labels AI-Generated Content, Even if Uploaders Don't Declare It

TikTok has recently announced that it will automatically label Artificial Intelligence Generated Content (AIGC) from other platforms, in collaboration with Adobe's CAI and C2PA (The Coalition for Content Provenance and Authenticity). This is done by reading the hidden metadata in AI-generated content.

According to the C2PA standard, when a user creates an image or video using AI, the file will contain C2PA-related metadata, including information on whether it was generated by AI and the platform used.

TikTok can directly read this metadata and label the content as AI-generated, even if the creator doesn't declare it as such when uploading the video.

Currently, to combat malicious users who intentionally deceive others with AI-generated content, TikTok requires creators to label their AI-generated content. This way, users can see the declaration when browsing videos.

Now, by reading the C2PA-related metadata, this process can be automated, and the AI-generated label will be added regardless of whether the creator labels it or not, unless the creator manually modifies the image or video to remove the C2PA metadata.

TikTok is the first video-sharing platform to implement automatic labeling of AIGC content by reading metadata. The platform plans to embed C2PA metadata into videos in the next few months, allowing the labels to be retained even if the videos are downloaded from TikTok.

Additionally, Adobe, Microsoft, and OpenAI are also members of the C2PA Alliance, and it is expected that more AIGC content will be added with C2PA metadata in the future.