YouTube tests AI creator avatars

YouTube is rolling out a Shorts tool that lets creators generate videos with an artificial version of their own face and voice. (9to5google.com) The setup starts with a “live selfie” in the YouTube app or YouTube Create app, where a creator records their face and reads prompts so YouTube can model their likeness and voice. Each generated clip can run up to eight seconds, and creators can stitch multiple clips into a longer Short. (9to5google.com) YouTube Chief Executive Officer Neal Mohan previewed the feature on January 21, 2026, saying creators would soon be able to make Shorts using their own likeness. By April 9, outlets tracking the rollout reported the tool was going live gradually. (techcrunch.com, engadget.com) YouTube has paired the avatar launch with disclosure rules it began building earlier. The company said in March 2024 that realistic altered or synthetic content must be disclosed, and its Help Center says Shorts made with YouTube’s own generative artificial intelligence tools are labeled automatically. (blog.youtube, support.google.com) The labels are not just text in a description box. YouTube says some altered or synthetic videos get a front-of-player notice, and its “How this content was made” system can add labels proactively when a creator does not disclose content that could mislead viewers. (blog.youtube, support.google.com) Reports on the new avatar tool say every avatar clip will also carry visible watermarks and machine-readable markers including SynthID and Coalition for Content Provenance and Authenticity credentials. That gives YouTube a second layer of disclosure beyond the standard synthetic-media label. (tubefilter.com, dataconomy.com) The legal questions start with consent and advertising. The Federal Trade Commission says endorsements must be truthful and not misleading, and influencers must clearly disclose material ties to brands, which means an artificial avatar pitching a product still falls under the same ad rules as the human creator behind it. (ftc.gov, ftc.gov) The second set of questions is about impersonation and voice cloning. The Federal Trade Commission has separately warned that artificial intelligence voice cloning can fuel fraud and said in 2024 that it was seeking stronger protections against artificial intelligence impersonation of individuals. (ftc.gov, ftc.gov) State publicity-rights law is moving at the same time. California updated its postmortem right-of-publicity statute in 2024 to cover some unauthorized digital replicas of a deceased personality’s voice or likeness, adding another legal backdrop for platforms and brands using synthetic performers. (california.public.law) YouTube has framed the broader push as both creation and protection. In November 2024, the company said it was developing likeness-management tools to help creators and artists control how their face and voice are represented on the platform. (blog.youtube) That leaves the new Shorts avatars sitting between convenience and verification. YouTube is making it easier to publish without filming, while also building a record inside the video that tells viewers the person on screen may be a synthetic copy. (9to5google.com, support.google.com)