YouTube rolls out AI clone patrol to spot deepfakes

Just a heads up, if you buy something through our links, we may get a small share of the sale. It’s one of the ways we keep the lights on here. Click here for more.

YouTube is finally giving creators a shield against their digital doppelgängers. 

Starting today, select members of the YouTube Partner Program can access a new AI detection tool that flags videos using their likeness, whether it’s a deepfake rant, a synthetic makeup tutorial, or an “AI reacts to AI” mashup gone wrong.

Here’s how it works: once a creator verifies their identity, they can head to the new Content Detection tab in YouTube Studio. 

There, YouTube will surface videos it thinks might feature their face or likeness. 

If something looks suspicious, say, a video where you’re reviewing a product you’ve never heard of, creators can submit a removal request directly through the platform.

The feature is rolling out slowly, with a first wave of lucky creators getting notified by email this morning. 

YouTube says it’ll expand access “over the next few months.” 

But in true beta fashion, YouTube warns that early users might see false positives, like clips from their own channels. 

It’s basically Content ID for faces, only this time the infringement is your actual existence.

This isn’t coming out of nowhere. YouTube first teased this AI likeness detection system last year, testing it with big-name talent from Creative Artists Agency (CAA), meaning a handful of Hollywood types got to try it before the rest of us. 

YouTube described the early version as a way to help “influential figures identify and manage AI-generated content that features their likeness at scale.” No more AI versions of The Rock selling crypto.

YouTube’s parent company, Google, has been walking a tightrope with AI, building new video tools while also scrambling to keep the chaos in check. 

Earlier this year, YouTube started requiring creators to label AI-generated content and cracked down on AI-cloned music mimicking real artists.

Now, the platform’s latest tool feels like a logical next step, part identity protection, part digital bouncer. Whether it’s enough to stop the flood of deepfake nonsense, though, is anyone’s guess.