TikTok is the first to watermark AI content


A watermark should make it clear to users whether they are watching a real TikTok video or something generated by AI.

A first for TikTok: the short film app is the first social network that automatically generates AI-generated content from a watermark will provide. To do that, it partners with the Coalition for Content Provenance and Authenticity (C2PA).

Metadata for AI

At C2PA, technology was developed that immediately lets you know whether something comes from AI or not. In fact, it is a piece of metadata: that is extra information that is included with a photo or video file. Currently, this metadata is mainly used to immediately state the time and location. Now it contains an extra piece of information that lets you know whether something is AI-generated or not.

C2PA is a partnership between Arm, BBC, Intel, Microsoft, Adobe, Microsoft and Truepic. With their ‘content credentials’ they want to develop a royalty-free standard to warn people about AI content. The fact that TikTok is now the first social medium to use this standard is good news.

You will not see those ‘content credentials’ yourself. Based on this information, TikTok can determine who made a certain video, whether it was edited and whether AI was involved. The social media app uses that information to determine which videos get a watermark.

TikTok is already placing labels on videos that contain AI effects. The company itself will also start attaching content credentials to TikTok videos. That data remains attached to the video when you download it, so that other websites can also warn about AI use.

Not a foolproof solution

The question is whether that is enough. Anyone with a little knowledge could remove the metadata that TikTok looks at. The app further encourages users to label their own content if it involves AI. The medium naturally relies on its users for this, who must be willing to cooperate. In other words: the fact that TikTok supports the new form of metadata is good news, but does not in any way mean that the battle against AI-generated content is over. That will be an ongoing battle, not only for TikTok but also for other social media.


Featured article

AI

White House is pushing AI players for more transparency and watermarking