The short video platform TikTok is launching a test of a new tool called Likeness Detection among a limited group of American creators. This feature is designed to detect AI-generated content that uses a specific creator's face. The system allows users to find deepfakes or fake accounts and submit complaints directly through the platform's interface.
To activate the tool, creators must undergo mandatory identity verification through a third-party service called Jumio. The process involves creating a real-time selfie and scanning an identification document. It is important to note that TikTok states a policy of not storing copies of documents — biometric data is used solely to find matches with potential deepfakes, not for long-term storage in the platform's database.
How does it work?
The Likeness Detection mechanism analyzes the visual and behavioral characteristics of a creator's face, comparing them with content uploaded to the platform. When matches with AI-generated materials are detected, the system automatically flags such videos and provides the creator with the ability to submit a complaint. This is particularly relevant given the growing number of deepfakes, which can cause reputational damage or be used for fraud.
From my perspective, this step by TikTok is timely but raises questions about the scale of implementation. Testing on a small group of creators from the US is only the first stage. The deepfake problem is becoming critical for an industry where visual content is the primary asset. However, the system's effectiveness will depend on the accuracy of the algorithms and the speed of response to complaints. Additionally, the use of the third-party verifier Jumio could become a vulnerability point for data privacy, despite the platform's promises.
Analytical commentary: The introduction of Likeness Detection is not only a tool for protecting creators but also a precedent for the entire social media industry. In the coming years, we will see similar solutions from other platforms, as the regulation of AI-generated content becomes inevitable. However, the main challenge is the balance between privacy protection and the effectiveness of combating deepfakes.