Meta built an AI detection tool to ID images and video created with its new models

Editor's take

Meta has developed a tool capable of identifying AI-generated images and videos produced by its own models, such as those from the Make-A-Video and Make-A-Scene projects. This move signifies a proactive, albeit potentially incomplete, step towards addressing the growing challenge of synthetic media authenticity.

The significance lies in Meta's attempt to provide a technical countermeasure against deepfakes and misinformation originating from its platforms, a critical concern as generative AI capabilities become more accessible. However, the existence of rate limits on this detection tool raises questions about its practical efficacy and Meta's commitment to widespread, unhindered deployment for all users.

Future developments will hinge on the tool's accuracy against a wider array of generative models beyond Meta's own, and whether these rate limits are a temporary measure or a permanent constraint. The true test will be whether this detector can keep pace with the rapid evolution of AI generation techniques and its integration into the broader content ecosystem.