Google's Gemini Omni lets you video-clone yourself with AI avatars
Clone yourself for videos, but trust concerns loom with Google's new Omni tool.
Google unveiled Gemini Omni at I/O 2026, a multimodal AI video generator that aims to revolutionize video creation much like Nano Banana did for images. Omni can accept any combination of text, images, audio, and video inputs to produce high-quality videos grounded in Gemini's real-world knowledge. The tool includes an avatar feature that lets users create a digital clone of themselves, complete with their voice and appearance, enabling scripted video generation without actual recording.
Google emphasizes responsibility by incorporating SynthID digital fingerprinting to mark Omni-generated content, but acknowledges editing audio and speech in videos is still under testing. Additionally, Omni boasts an improved physics model that understands gravity, kinetic energy, and fluid dynamics, making generated scenes more realistic. The first tier, Gemini Omni Flash, rolls out today to the Gemini app, Google Flow, and YouTube Shorts. While promising for creators, the tool stirs debate about AI slop and trust in video authenticity.
- Gemini Omni combines text, images, audio, and video inputs to generate videos with real-world reasoning.
- Avatars let users create a digital twin that looks and sounds like them, with SynthID watermarking for verification.
- Physics model simulates gravity, kinetic energy, and fluid dynamics for more realistic motion.
Why It Matters
Gemini Omni could democratize video creation but also amplify AI-generated disinformation, demanding new verification standards.