TikTok adopts a new tool to better identify content created with artificial intelligence on the social network. It is in addition to the one it already uses for transparency purposes.

Since the massive rise of generative artificial intelligence tools, it has never been easier to make realistic photos, images, sounds or videos. This is not necessarily a problem in itself, but it becomes a problem when creation is presented as true. Of course, it is theoretically possible to recognize certain AI contents with the naked eye, except that it is not always obvious. And when they are disseminated on a large scale via social networks, the impact can be enormous.

On TikTok, for example, fake news from AI is increasingly present. Aware that this technology "offers incredible creative opportunities, but can confuse or mislead viewers [...]," the Chinese platform launched last year a clear identification system for all content made with the artificial intelligence tools it offers. However, this does not prevent anyone from uploading an AI video created with, for example, Sora from OpenAI and not specifying anything. This point will change.

TikTok announces the automatic implementation of a specific symbol: Content Crédentials (cr) developed by the Coalition for Content Provenance and Authenticity (CP2A). It may not be unknown to you, it is the one that Adobe unveiled last year with the aim of making it a universal label. Microsoft has already adopted it for example. This system adds metadata to the content, which is used to clearly recognize and identify those created by artificial intelligence.

The "cr" logo is to images and videos now, while audio-only content will benefit "soon," without being more precise. TikTok, still threatened with being banned from the United States, wants to be as transparent as possible, along with Facebook, Instagram and Google. has its own way of identifying content generated by AI, but China's social network is the first video sharing platform to implement CP2A technology.

TikTok: the content generated by AI will be much easier to identify, here's how