Meta plans to make it easier to tell if content on Threads, Instagram and Facebook was created with artificial intelligence. To do this, the company says it will start tagging such content “in the coming months”.
Today, Meta announced that it is working on the ability to detect AI-generated content. When ready, this capability will allow the company to add tags to content submitted to Facebook, Instagram and Threads. The tags will also be available in all languages supported by each app.
Once the capability is deployed, it will affect images on Meta’s platforms. For this, it will identify industry standards used to identify AI images, such as invisible watermarks, embedded metadata, visible markers. If it finds that an image does not carry these signals, Meta notes that such images may go unnoticed.
This capability will not be limited to identifying images created with Meta’s AI tools. The company also aims to be able to identify AI images created with AI tools from other companies such as Google, OpenAI and Adobe.
It’s not just images that Meta is focused on; the company also wants to tag AI-generated video and audio. However, as Meta notes, these signals are not present to the same extent in AI tools that produce video and audio. Until those signals are found, Meta says that for now, it will “add a feature that allows people to annotate when they share AI-generated video or audio.”
Labeling videos or audio as AI-generated will not be optional. Meta says it will require users to “use this annotation and labeling tool when they share a realistic-looking video or realistic audio that has been digitally generated or modified.” The company adds that it may penalize users if they find unlabeled AI-generated videos or audio.
