Google DeepMind unveiled Veo 3, their new video generation model that for the first time can produce synchronized audio with visuals, including traffic noises, birds singing, and character dialogue. The company also introduced Flow, an AI filmmaking tool designed to create cinematic content using Veo, Imagen, and Gemini models together. These tools aim to give creators unprecedented control over their visual storytelling.

Veo 2 has been enhanced with reference-powered video capabilities, camera controls, outpainting, and object add/remove features. Meanwhile, Imagen 4 delivers stunning image quality with improved typography and fine details, supporting aspect ratios and 2k resolution. These updates expand creative possibilities while maintaining consistency in generated content across different platforms.

Lyria 2 powers Music AI Sandbox and YouTube Shorts, offering musicians composition tools and real-time interactive generation. DeepMind has watermarked over 10 billion images with SynthID and launched SynthID Detector to verify AI-generated content. The company emphasizes responsible development through partnerships with creative industries who helped shape these tools.