Google previews Genie 4 with audio and real-time scene editing
The AMW Read
Audio, live scene editing, and longer simulations meaningfully extend Google's interactive-media offering, but limited access and unresolved reliability constrain demonstrated impact.
Google previews Genie 4 with audio and real-time scene editing
Google DeepMind released Genie 4 in research preview for trusted testers on October 27. The world model adds synchronized audio, more realistic visuals, and text prompts that change scenes during a simulation. Users can alter weather, adjust camera angles, or add people and objects. Google says scenes can remain consistent for up to two minutes, twice its predecessor's limit. Access through Project Genie for Google AI Ultra subscribers and selected Gemini API and Google AI Studio partners is planned in the coming weeks.
The release places Google in the generative-media contest over interactive creation, where maintaining a scene through user actions matters alongside image quality. Runway's GWM Worlds 2 also combines playable worlds with audio. Google is testing a complementary role for Gemini: planning steps and checking frames while Genie generates the simulation. Trusted testers are exploring film prototyping and storyboarding, potentially giving Genie a role before finished video production. That workflow remains exploratory; Google says the complete product experience is still unclear.
Builders should test whether Genie preserves objects, scene details, and useful audio across repeated edits before making it part of a production workflow. Google acknowledges visual drift, audio accuracy problems, and illegible text. Its Gemini Robotics experiment suggests a possible simulation tool for testing robot scenarios, but the report provides no evidence of validated real-world transfer. For investors, the concrete question is whether live editing becomes a repeatable creative workflow beyond an impressive research preview.
