The short video social platform TikTok has launched a pilot test of a new tool called Likeness Detection. Currently, the feature is available to a limited group of creators in the US. Based on my market analysis, this is a step in the right direction for a platform that has long remained vulnerable to attacks using synthesized faces.
How does deepfake protection work?
The tool is designed to automatically detect content created using artificial intelligence that features a specific blogger's face. The system scans the platform for suspicious deepfakes and fake accounts. When a match is found, the creator has the opportunity to file a complaint about the violation.
The key activation step is mandatory identity verification through a third-party service, Jumio. The procedure involves two steps: a real-time selfie and scanning an identification document. The company guarantees that it does not store copies of documents, and biometric data is used solely to search for matches on the platform.
Expert assessment
From a technical standpoint, implementing via Jumio is a reasonable compromise between security and privacy. However, it is worth noting that the effectiveness of Likeness Detection will directly depend on the quality of training data and the algorithm's ability to distinguish legitimate content from malicious deepfakes. In its current configuration, the tool only works based on creator submissions, which leaves loopholes for professional forgers. In my opinion, this is merely the first step in a long arms race between platforms and AI content generators.