Tikok can create short videos with AI from given pictures

The video portal Tiktok introduces a new function to create moving images. “Ai Alive” allows users to produce short videos with the support of artificial intelligence (AI) from static pictures. The AI ​​should adhere to the guidelines that are given to it by text. Tikok promises to check the videos created by AI in advance and mark them accordingly in order to be able to identify them as AI content.

This should be integrated into the videos via C2PA metadata. Quite exactly a year ago, Tiktok on the part of the Content Authenticity Initiative (CAI) and has been using C2PA identification for AI content since then. C2PA is an open standard to digitally label media and to verify the origin. Other AI services are also part of this initiative. Openai C2PA metadata integrates into generated images and Meta also uses C2PA for its AI image generator.

AI videos only in Tikok stories

At TikTok, the AI ​​function can only be reached via the “Story Camera”. The user selects a single photo from the story album, whereupon an icon for AI alive appears. According to Tikkok “Use Ai Alive Intelligent Processing tools that enable everyone, regardless of their processing experience, to convert static images into captivating short videos that are enriched with movement, atmospheric and creative effects”.

If Ai alive is activated after selecting a photo, a text input field appears that is pre -fulfilled with the words “awakening this photo to life”, how The Verge writes. Accordingly, the process takes a few minutes, although the generated video is only a few seconds. In addition, the AI ​​ignored some commands or simply misunderstood them, for example if a cat shown and the video style video should be created.

Tikkok checks AI videos several times

The somewhat longer creation of the AI ​​video could be due to the security measures of Tiktok. Because every uploaded image, the textual instructions and the generated videos are first checked whether they violate Tikok guidelines. One last exam takes place when the user publishes the AI ​​generated video as a story. In addition, other users can report AI videos if they suspect violations of the guidelines like any content on the platform.