The battle against synthetic impersonation is shifting toward the protection of digital identity. TikTok has begun testing a new opt-in tool designed to scan the platform for AI-generated content that uses creators' likenesses without their permission. This initiative aims to give authors a direct mechanism to identify and report deepfakes that compromise their image.

Identity Verification and Data Privacy

To access this feature, creators involved in the test — currently limited to a select group of users in the US — must undergo a rigorous authentication process. The procedure is handled via Jumio, a third-party service requiring an ID check and a real-time selfie scan. According to TikTok US spokesperson Zachary Kizer, the company does not retain ID documents or facial information used for verification, emphasizing a privacy-first approach despite the sensitive nature of the data.

A Multi-Layered Defense Strategy

The likeness detection system is not an isolated intervention but part of a broader strategy to combat misinformation and synthetic spam. TikTok is implementing several security measures:

  • AI Spam Mitigation: The platform is testing enhanced systems to target accounts dedicated solely to mass-producing AI content, particularly regarding politics and current events.
  • Provenance Standards: To improve transparency, TikTok supports C2PA Content Credentials, cryptographically signed metadata that allows the origin of a media file to be read automatically upon upload.
  • AI Literacy: An educational hub and an in-app guide are being launched to teach users how to distinguish real content from synthetic versions.

The Platform Race

TikTok is not alone in this challenge. YouTube has already deployed a similar technology, making it available to all adult users and alerting them when their face appears in AI-generated videos. Both companies are struggling with a structural issue: while proprietary labeling systems (which have already flagged 3 billion videos on TikTok) work for internal tools, they remain vulnerable to files uploaded from external platforms.

This evolution reflects a global trend toward the formalization of digital identity, echoing proposals like DNSid for AI agent identity to ensure that every synthetic entity is traceable and securely verifiable.