The short video platform TikTok has begun limited testing of a new tool called Likeness Detection. Currently, the feature is only available to a small group of creators in the United States.
The essence of the development is the automatic search for content created by neural networks that uses a specific creator's face. If the system detects a deepfake or a fake account, the blogger can promptly file a complaint. This is a step in combating the growing threat of digital forgeries and fraud using well-known faces.
To activate the tool, users must undergo a two-step identity verification through the Jumio service. The procedure includes a real-time selfie and scanning an identification document. The company emphasizes that it does not store copies of documents, and biometric data is used solely to find matches with videos on the platform.
This is an important precedent for the entire industry. As generative models improve in quality, distinguishing real videos from synthesized ones becomes increasingly difficult. Such measures, even on a voluntary basis, set a precedent for implementing mandatory verification of creators whose faces may be used without their consent.
In my opinion, the success of Likeness Detection will depend on the speed of the algorithms and their ability to distinguish legitimate parodies from malicious fakes. If the system proves its effectiveness, we will likely see similar solutions on other major platforms in the coming quarters.