Google’s Omni Model Promises ‘Anything-to-Anything’ AI, but the Reality is Still Uncanny

Google’s new Omni model aims to transform any input into any output. We put its video generation and deepfake capabilities to the test.