Google’s Omni AI lets you video-clone yourself, raising intrigue and concern

Google has unveiled Gemini Omni, a new AI video tool that promises to transform how creators produce content,though opinions are sharply divided on whether this is a creative breakthrough or a floodgate for low-quality AI-generated videos. Announced today, the tool represents what Google calls a leap comparable to Nano Banana’s impact on image generation, but now applied to video.

The company describes Omni as “where Gemini’s ability to reason meets the ability to create.” Users can combine images, audio, video, and text as inputs to generate high-quality videos grounded in Gemini’s real-world understanding. The rollout begins today, starting with Gemini Omni Flash, and will extend to the Gemini app, Google Flow, and YouTube Shorts. It remains unclear whether the web version of Gemini will support Omni directly or require the Flow interface.

Among the most intriguing,and controversial,features is the ability to create a digital avatar of yourself. Google says you can generate videos “with your own voice by using Avatars, which create a digital version of yourself so you can look and sound like you.” For a regular YouTube creator like myself, this raises immediate questions: Could I feed a script into my digital twin and let RoboDave handle a bad hair day? Would my audience notice or care? And more importantly, would I lose the practice and training that comes from actually performing on camera?

Google is addressing authenticity concerns by embedding SynthID digital fingerprinting into Omni-generated videos, allowing them to be verified as AI-produced. The company also notes it is “still working to test” how to responsibly handle editing audio and speech within videos.

Another standout upgrade is the inclusion of a physics model in video generation. Google says Omni has “an improved intuitive understanding of forces like gravity, kinetic energy, and fluid dynamics,” moving beyond simple pattern matching. This means generated videos should behave more realistically,objects falling, colliding, or reacting to forces in a way that mimics the physical world. The tool can also build detailed explainers from short prompts, leveraging Gemini’s knowledge to break down complex ideas. I’ve already seen NotebookLM produce explainer videos from marketing documents that were better and faster than anything I could create manually, though the visuals were rough. If Omni inherits that reasoning capability, the potential is enormous.

Input variety is another major leap. While Nano Banana could recontextualize a single image, Omni can take text, images, video, or voice recordings and produce a cohesive video output. Currently, only voice recordings are accepted as audio input, but Google says other audio types will roll out soon. Users can create scenes, match styles, describe what they want in natural language, and maintain character consistency throughout the video.

Perhaps most appealing to anyone who has suffered through video editing is the promise of conversational editing. Google explains that “every instruction builds on the last. Your characters stay consistent, the physics hold up, and the scene remembers what came before.” You can change elements within a video,remove obstructions, alter backgrounds, or swap objects,using natural language commands. The company says you can “change specific things, or change everything. Your video becomes the starting point for something you never could have filmed yourself.”

However, critical details remain unclear. Google hasn’t specified supported video formats or resolutions. Will Omni handle professional 16:9 4K or 8K footage, or is it primarily designed for the YouTube Shorts ecosystem? When OpenAI launched Sora, it was a novelty that rarely fit into professional workflows. I’m hoping Omni integrates with tools like Final Cut Pro, Premiere Pro, or DaVinci Resolve, either natively or through an API. Google says Omni’s features will roll out to enterprise customers and developers via a Google API, which suggests professional integration is possible.

Another open question is watermarking. Will Omni embed the diamond watermark seen on Nano Banana images? Such marks help identify AI-generated content but can interfere with professional use. Whether Google offers licensing tiers to remove the watermark,or third-party tools emerge to bypass it,remains to be seen.

Would you trust a digital twin to deliver your next video? Share your thoughts in the comments.