Google AI Updates: July 20, 2026

1. Google DeepMind Repurposes a Video Generator for Computer Vision

Google DeepMind. Google DeepMind published GenCeption, a system that adapts a pretrained video generator for computer-vision tasks such as depth estimation and segmentation, matching or beating specialized state-of-the-art models while using 7 to 500 times less training data. Trained largely on 7,500 synthetic videos, it generalized to real footage and unseen categories, which the researchers argue supports the contested claim that video generators already encode the world models computer vision has lacked. The work suggests large video models could become foundation models for vision the way language models did for text. Source