TikTok announced its plan to introduce labeling for AI-generated images and videos uploaded to its platform, utilizing a digital watermark called Content Credentials. This move aims to address concerns raised by researchers about the potential misuse of AI-generated content during upcoming US elections. TikTok, along with a consortium of tech companies, previously committed to combating this issue.
While TikTok already labels AI-generated content produced within its app, the new initiative extends this labeling to content generated externally. Adam Presser, TikTok’s head of operations and trust and safety, emphasized that they have strict policies against unlabeled realistic AI content and will remove any violating content.
Content Credentials, developed by the Coalition for Content Provenance and Authenticity (which includes Adobe and Microsoft), has been adopted by platforms like YouTube and Meta Platforms (owner of Instagram and Facebook). This system requires cooperation between the AI tool provider and the platform to implement industry standards effectively.
For instance, when an image is generated using OpenAI’s Dall-E tool, OpenAI adds a watermark and data to the file to indicate its origin. If such a marked image is uploaded to TikTok, it will be automatically flagged as AI-generated.
TikTok, owned by China’s ByteDance, faces regulatory challenges in the US, with a recent law requiring ByteDance to divest TikTok or face a ban. TikTok and ByteDance have taken legal action, arguing that the law violates the First Amendment.
