TikTok quietly tests a tool to hunt deepfake doppelgangers

TikTok is wading into the deepfake arms race with a new opt-in tool that lets creators scan the platform for unauthorized AI-generated versions of themselves. The feature, spotted by social media consultant Matt Navarra, is currently in a limited test with a small group of US-based creators, according to company spokesperson Zachary Kizer.

The process isn’t frictionless. Creators who want to participate first have to verify their identity through Jumio, a third-party firm that handles real-time selfie scans and ID checks. Kizer emphasized that TikTok won’t hang onto those documents, and facial data is used strictly for matching and flagging potential impersonation — not for training AI models or building a biometric database. After passing that gate, TikTok’s systems crawl the platform looking for AI-generated content that appears to be using the verified creator’s face without permission.

From there, it’s a review-and-report workflow. Creators can see what TikTok’s detection systems surfaced and choose to report posts or entire accounts they believe are violating their likeness. It’s a reactive approach — find the fakes after they’re already live — rather than a pre-filter that blocks them at upload. That’s the same basic playbook YouTube recently adopted when it rolled out similar likeness detection to all adult users, though Google’s video giant had been working on its version for well over a year before going wide.

The timing isn’t random. AI-generated celebrity and creator deepfakes have exploded across short-form video platforms in the past eighteen months, from fake MrBeast giveaways to unauthorized Taylor Swift endorsements. TikTok’s tool gives individual creators a lever to pull, but the burden is still on them to opt in, verify, and manually act on what the system finds. For a platform that thrives on viral remixing and impersonation, the line between parody and theft is about to get tested in courtrooms and comment sections alike.