Google Vids now lets you star in your own AI videos

Google is betting that even after OpenAI benched Sora, people still want to put themselves into AI-generated video. On Thursday, the company announced a significant update to Google Vids, its Workspace video tool, that includes the ability to create a custom digital avatar from a selfie and a voice recording. It’s a direct pivot from a staid workplace presentation helper to something that looks a lot more like a direct competitor to specialized AI avatar startups like HeyGen and Synthesia.

The engine behind the new tricks is Gemini Omni, Google’s multimodal AI model. Instead of just text-to-video, Omni lets you mix a written prompt with reference images you upload, blending them into the final clip. It’s also handling practical production tasks like background swaps, lighting fixes for phone footage, and applying effects. Crucially, Omni now supports step-by-step edits, so you can tweak a video iteratively without trashing the whole project and starting from scratch every time.

By keeping Vids firmly inside Google Workspace, the company is signaling that this is a tool for company updates, training videos, and sales pitches—not a playground for viral weirdness. The corporate guardrails are pretty clear. Avatars are tied to the account holder’s likeness and Google account, and they’re invisibly watermarked with SynthID. Access is also gated to users 18 and older in certain regions, which feels like a direct lesson learned from watching OpenAI let users generate clips of Sam Altman with Sora.

Whether anyone actually needs a deepfake of themselves for a quarterly business review is an open question. The line between a productivity tool and a novelty is thin here. But for companies already paying for Workspace and looking to scale video content without a production crew, the ability to generate a talking-head video by typing a few sentences is genuinely utilitarian. The real test is whether the avatars look convincing enough that employees don’t cringe when they see themselves on screen.