TikTok announced it is starting to automatically label AI-generated content (AiGC) coming from other platforms. The company has teamed up with Adobe"s Content Authenticity Initiative (CAI) and Coalition for Content Prevenance and Authenticity (C2PA), which includes Adobe and Microsoft among its founding members.
The ByteDance-owned company already labels AI-generated content created on its platform such as content made with TikTok"s AI effects. It requires creators to label realistic-looking AIGC and over 37 million creators have used the tool that gets the job done.
"While most people want to enjoy AI-generated content responsibly, there will be bad actors who use AIGC to intentionally deceive others," the company said in a blog post explaining its latest move.
TikTok will identify AI-generated content with the help of Content Credentials - a technology developed by C2PA that makes it easier to spot AIGC by adding the required metadata to digital content. Content Credentials has been implemented in other tools and services such as Adobe"s Firefly, Photoshop, and images generated by Azure OpenAI DALL-E models.
TikTok has started rolling out the ability to read Content Credentials on images and videos, with support for audio-only content coming soon. It"s the first video-sharing platform to implement the technology, and over the coming months, it will start attaching Content Credentials to TikTok content, which will stick to it even if the content is downloaded.
TikTok will be able to label more AI content as other platforms adopt the technology. "This means that the increase in auto-labeled AIGC on TikTok may be gradual at first since it needs to have the Content Credentials metadata for us to identify and label it," it said.
Anyone can use C2PA"s Verify tool to spot "AIGC that was made on TikTok and even learn when, where and how the content was made or edited."