YouTube just made a Shorts deepfake machine so creators don’t have to be in their own videos

Hey YouTubers! Do you want to be rid of the pesky chore of actually appearing in your own videos? Well, good news: YouTube‘s latest AI feature is just for you!

A few months ago we covered that Google, which is absolutely determined to make fat dolla dolla bills before the gen AI bubble bursts, was experimenting with a fan-facing feature called Portraits. That feature, it said, would “let viewers conversationally interact with AI representations of participating creators.”

Basically, creators would no longer have to endure engaging with their adoring fans. Because who would want that?

Now YouTube is taking things a step further. Its next big idea is to have creators deepfake themselves so they never have to be on camera again.

Instead, their “AI avatars” will host videos for them.

“An avatar builds on existing ingredients-to-video features in YouTube’s creation tools, making it easier and more accurate to add yourself into your videos,” the platform explains in a YouTube Help Center post. “Avatars create a digital version of yourself so you can generate videos that look and sound like you, safely and securely.”

This feature is already rolling out to creators 18 and older around the world (except in Europe, possibly due to its more stringent regulations on both data privacy and generative AI). To make an avatar of themself, a creator has to enter the “AI Playground” section of YouTube’s app, then record a “live selfie” that will serve as training material for replicating their face and voice.

Once the avatar is done, creators can make Shorts

with it or inject it into other people’s videos via Shorts’ Remix feature. All they have to do is type in a prompt explaining what they want the avatar to do and say. No one else will be able to use the creator’s avatar in their videos, and the creator can delete it at any time.

YouTube–which, by the way, is already under fire for the amount of AI slop it spits out at viewers—told 9to5Google that AI avatars “gives users an easier way to include themselves safely and securely in videos.”

Yes, we suppose it is easier to type a few words in a prompt box. But are viewers actually going to be into videos that are the YouTuber equivalent of ChatGPT holding up a life-size cardboard Mandalorian cutout, wiggling it around, and pretending to be Pedro Pascal?

Either way, they’ll at least know when a creator has used their deepfake instead of doing the work: YouTube says all videos generated with avatars will have YouTube’s standard AI disclosure, plus will “include visible watermarks and digital labels like SynthID and C2PA to disclose AI-generated content.”

9to5Google reports AI avatar clips are restricted to eight seconds long, but there’s no limit to how many individual eight-second clips creators can record and then sew together into one longer video.

What else is there to say here? For a platform that keeps insisting creators are its primary concern and AI should amplify human creativity, YouTube is sure bent on giving creators ways to remove themselves from the creative process–not to mention ways to automate their connections with fans.

Do creators actually want this? Do viewers? We’ll be keeping an eye on feedback as this feature goes out to more channels in the coming weeks.