Meta is preparing a tool to detect AI content

As AI-generated content floods the web, reaching over 50% according to recent observations, tech giants are beginning to adapt.

Having largely contributed to democratizing these uses, some players are now seeking to better regulate them.

It is in this context that Meta is reportedly working on a still discreet feature, but one that reveals the current issues surrounding generative AI…

An AI detector integrated directly into Meta AI

According to some users, a tool called “AI Detector” has been spotted in the Meta AI interface. This feature, still under development, would allow users to analyze content to determine if it was generated by artificial intelligence. The first traces were identified by TestingCatalog through the activation of internal features in the application's code, a common practice for discovering unannounced tools. Currently, the detector is not active, and each click on the option leads to an inactive page, indicating that the project is still in the testing phase. However, based on available information, this tool may initially focus on text analysis before expanding to other formats such as images, audio, or video. A development consistent with Meta's massive investments in multimedia content generation…

A response to the proliferation of synthetic content?

This development comes in a particular context, where in recent months, Meta has multiplied the generative AI features on its platforms, such as the recent transformation of photos into stories. Unfortunately, this has resulted in an explosion of what some are now calling "AI slop," mass-produced, often low-quality synthetic content.

Indeed, the group is even experimenting with new formats, such as feeds where AI-generated content occupies a central place. This approach profoundly transforms the user experience, but also raises questions about the reliability and readability of the information disseminated.

Faced with this situation, the development of an AI detector appears as a form of defensive response.

But this raises a significant question: will this tool be able to identify content from any model, or only content produced by the Meta ecosystem? In practice, this choice will directly impact its reach. Furthermore, while no official announcement has been made at this stage, this initiative illustrates a fundamental trend: after accelerating content production via AI, platforms are now seeking to better identify it, and potentially regain control over it.