Google bets on avatars

YouTube is starting to let creators put an artificial version of themselves into Shorts, using a verified face and voice instead of a camera shoot. The rollout began on April 9, 2026, and YouTube says it will happen gradually rather than all at once. (engadget.com) The pitch is simple: record a live selfie, give YouTube your voice, and the app can generate short videos where your avatar speaks for you. Early reports say the feature is available globally except in Europe, which hints at the regulatory line Google is trying not to cross first. (androidheadlines.com) This did not come out of nowhere. On January 21, 2026, YouTube chief executive Neal Mohan said creators would soon be able to make Shorts with their “own likeness,” folding synthetic selves into the same product push that already included artificial intelligence clips, dubbing, and editing tools. (blog.youtube) YouTube has been building the guardrails first and the avatar product second. In 2024 it announced “likeness management” tools for faces and voices, and by March 2026 it had expanded its likeness detection pilot to journalists, government officials, and political candidates who want to find artificial intelligence copies of themselves on the platform. (blog.youtube, blog.youtube) The company is also making those controls visible to ordinary creators. YouTube’s help pages say likeness detection lets eligible users review videos where their face appears altered or generated by artificial intelligence and then choose whether to request removal through the privacy complaint process. (support.google.com) That is the key difference between Google’s move and the first wave of consumer artificial intelligence video hype. OpenAI’s Sora web and app experiences are being shut down on April 26, 2026, with the application programming interface scheduled to end on September 24, 2026, even as YouTube is pushing synthetic video deeper into a product used by billions. (help.openai.com, blog.youtube) The economics help explain the split. A standalone video generator has to win users one prompt at a time, while YouTube can drop avatars into Shorts, where the platform already says the format averages more than 200 billion daily views and sits inside the same ad, recommendation, and creator payment machine. (blog.youtube) Google is also not treating avatars as a separate bet. At its 2025 Made on YouTube event, the company said it was integrating Google DeepMind’s Veo 3 Fast into Shorts for generated backgrounds and clips with sound, which means the avatar tool lands inside a stack that already makes synthetic scenes, synthetic edits, and synthetic speech. (blog.youtube, blog.youtube) So the fight is no longer over whether artificial intelligence video exists. The fight is over which company can wrap it in identity checks, removal tools, and a distribution system big enough that creators will trust it with their face. (support.google.com, blog.youtube) That is why this launch feels bigger than one new button in Shorts. Consumer video generation did not disappear when legal risk, cost, and deepfake fears showed up; it moved toward the company that already owns the audience, the moderation system, and the upload feed. (theverge.com, help.openai.com)