The deepfake arms race is heating up.
TikTok just dropped a hint that they’re fighting back. An opt-in detection tool is rolling out to some US creators. It scans for unauthorized AI-generated likenesses of your face. If you see yourself where you don’t belong, you can finally report it directly through the app.
Matt Navarra, a social media consultant with an eye for these things, spotted the test first. TikTok US spokesperson Zachary Kimer confirmed the news. But there’s a catch. It’s not for everyone, not yet. Just “some” US creators get access for now. YouTube has a similar setup already, open to all adult users who want it.
Why verify your face to catch deepfakes?
It sounds counterintuitive, right? Give a company your face so they can stop people from using it.
You have to jump through hoops to use the tool. First up: verification. You need a service called Jumio to prove who you are. This means a real-time selfie scan. Then, you scan a government-issued ID. It’s strict. It’s bureaucratic. But here is the important part: TikTok claims they don’t keep your ID docs. The facial data stays strictly for matching your likeness against potential violations.
TikTok does not retain ID documents. Facial info is used only to identify unauthorized use.
Still. Handing over biometric data makes some people twitchy. It’s a trade-off. Privacy for protection. You decide which risk feels heavier.
What actually happens after verification?
Once the trust is established, the machinery kicks in. The system starts looking. It scans for content that might be AI-generated using your likeness. It doesn’t just wait for someone to accuse you of impersonating another creator. It proactively finds potential matches.
You then see what they found. You have the final say.
If you see a post using your face in a way you didn’t authorize, you flag it. TikTok then reviews and acts. It closes the loop. Previously, reporting this stuff was a mess. Now it’s built into the workflow.
Which is better for protecting your digital identity, TikTok’s new scan or manual reporting? For those who have been targeted by realistic video forgeries, this looks like a faster lane to resolution.
Should you wait for wider release?
The rollout is slow. Why? Probably to test
