YouTube opens deepfake detector to Hollywood

If you're in the public eye, few things are more infuriating than waking up to find your face plastered across a shady AI-generated scam ad.

On Tuesday, YouTube announced that it is expanding its AI likeness detection tool to the entertainment industry, opening up access to celebrities, talent agencies, and management companies.

Spotting deepfakes without a channel

The expansion was developed with input from major talent powerhouses including CAA, UTA, WME, and Untitled Management. Crucially, performers and public figures don't even need to run an active YouTube channel to participate.

To enroll, talent must submit an ID along with a selfie video. The system then automatically scans YouTube for AI-generated content that matches the person's face. If an enrolled celebrity spots an unauthorized deepfake, they can submit a privacy violation claim, file a copyright takedown request, or simply leave it alone.

YouTube points out that takedowns are not guaranteed, because parodies and satire are still protected under existing platform rules. So far, the tool is strictly focused on faces, though YouTube says voice detection is on the roadmap. The company has also been lobbying for federal protections in Washington by backing the NO FAKES Act.

This marks the latest phase of a gradual rollout. YouTube began testing the feature with select creators last fall, expanded it to politicians and journalists in March, and is now bringing Hollywood into the fold. According to YouTube, creators who used the tool so far requested a very small number of removals.

Takedowns for now, monetization later?

YouTube itself compares the tech to Content ID, its long-running copyright system. But there is a glaring difference: Content ID lets rightsholders monetize unauthorized uploads instead of wiping them off the web.

Likeness detection doesn't have a revenue-sharing toggle yet. But the industry is clearly leaning that way. Talent agencies like CAA already maintain biometric databases of their talent, and YouTube recently introduced a tool allowing creators to digitally clone their likeness for videos.

Many talent managers see generative video as an inevitable medium for fan engagement. For some entertainers, unauthorized AI clips are a nuisance to be wiped out; for others, they could become a new revenue stream once payment pipelines are in place.

Why it matters

YouTube is taking the defensive playbook it built for music piracy and applying it to generative AI, which we think is a practical move. For public figures, automated face scanning is infinitely more realistic than hiring legal teams to manually scour millions of daily uploads for scam endorsements.

Right now, the value is almost purely defensive, helping public figures quickly shut down dangerous impersonations. But YouTube's refusal to auto-delete everything is going to put its review team to the test, especially when the line between a scam and acceptable satire gets blurry.

We think the real endgame here is monetization. If likeness detection eventually evolves into a Content ID-style ad split, it could effectively legalize and monetize deepfakes across the platform, turning a copyright headache into a brand-new licensing market for Hollywood.