{"slug": "googles-new-ai-world-model-can-generate-human-voices-not-just-video", "title": "Google’s New AI World Model Can Generate Human Voices, Not Just Video", "summary": "Google DeepMind launched Genie 4, the latest version of its AI world model, as a limited research preview for a small group of early testers, adding automatic audio generation and more fluid human motion and voices to the video simulation capabilities of Genie 3, which Google unveiled in August of last year. Google said in a Tuesday blog post that \"the most dramatic leap is how Genie 4 simulates people,\" and that \"understanding and simulating the world is a crucial aspect of building AGI.\" All Genie 4 videos and audio are tagged with Google's SynthID watermark, and Google plans a broader release starting with Google AI Ultra subscribers without specifying a date.", "body_md": "Google DeepMind has introduced the latest version of its AI world model, which according to the company “unlocks unprecedented realism” and specializes in simulating human beings.\n\nLaunched today as a limited research preview among a small group of early testers, the new model, called Genie 4, can generate photorealistic videos which users can modify in real-time via simple natural language prompts. Unlike its predecessor [Genie 3](https://deepmind.google/blog/genie-3-a-new-frontier-for-world-models/)—which Google first unveiled in August of last year—the new model can also generate audio. And it can do so automatically, to fit whatever visual scene a user has crafted. “Every world carries the ambient sound of its own setting, from wind in the trees to waves crashing at the beach,” Google wrote in a [blog post](https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/) published Tuesday. Google also highlights the new model’s ability to simulate real-world physics, creating scenes that respond dynamically to shifting camera angles, changes in lighting, and so on.\n\nBut the model’s real differentiator compared with earlier versions, according to Google, is its ability to simulate humans. Bodily motion and facial expressions are now more fluid and realistic, whereas Genie 3’s depictions tended to be more mechanical. “The most dramatic leap is how Genie 4 simulates people, creating characters that move naturally, respond organically to their environments, and can be interactively steered in real-time,” Google wrote in its blog post. That leap extends to the simulation of human voices, too: Users can prompt AI-generated characters to converse with one another and respond to events unfolding in a video.\n\n“World models” have become a major area of investment for frontier AI labs as they try to build systems that can respond to virtually any kind of input from limited training data. They should be able to learn by observing and simulate the actual mechanics of the physical world, rather than just improving their ability to coherently string words together. Some experts, most notably the French computer scientist and Turing Award-winner Yann LeCun, have suggested that world models—rather than LLMs—are the most likely route to the development of true artificial general intelligence. Google appears to share that view: “Understanding and simulating the world is a crucial aspect of building AGI,” the company wrote in its Tuesday blog post.\n\nGoogle says Genie 4 is being rolled out with “strict safety guardrails” akin to those that the company has put in place around its other AI tools, and that it was subject to “rigorous safety evaluations to prevent the generation of harmful or malicious content.” The model is still in its infancy, and Google says it still occasionally struggles to maintain visual consistency and audio accuracy, particularly when simulations get longer and more complex. All videos and audio generated by Genie 4 will be tagged with SynthID, Google’s invisible watermark, according to the blog post. By limiting the model’s initial availability exclusively to a cohort of testing partners, Google says it’s aiming to collect feedback and make any necessary safety upgrades before a more general release, starting with Google AI Ultra subscribers. (The blog post didn’t specify when that broader release would happen, and Google didn’t immediately respond to Gizmodo’s request for comment.)\n\nAmong that group of early testers are video game developers, who according to Google are actively experimenting with Genie 4 to quickly build working prototypes via natural language prompts—call it “vibe gaming.” It’s not the first time one of Google’s AI tools has threatened to upend the old guard of gaming: the January [launch](https://blog.google/innovation-and-ai/models-and-research/google-deepmind/project-genie/) of its Project Genie web app [reportedly](https://www.bloomberg.com/news/articles/2026-01-30/unity-video-game-stocks-fall-as-google-s-ai-tool-genie-sparks-fears) caused video game stocks to plummet.", "url": "https://wpnews.pro/news/googles-new-ai-world-model-can-generate-human-voices-not-just-video", "canonical_source": "https://gizmodo.com/googles-new-ai-world-model-can-generate-human-voices-not-just-video-2000831335", "published_at": "2026-10-27 20:45:14+00:00", "updated_at": "2026-10-27 21:12:19.578583+00:00", "lang": "en", "topics": ["artificial-intelligence", "generative-ai", "ai-research", "ai-products"], "entities": ["Google", "Google DeepMind", "Genie 4", "Genie 3", "SynthID", "Google AI Ultra", "Yann LeCun", "Project Genie"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/googles-new-ai-world-model-can-generate-human-voices-not-just-video", "markdown": "https://wpnews.pro/news/googles-new-ai-world-model-can-generate-human-voices-not-just-video.md", "text": "https://wpnews.pro/news/googles-new-ai-world-model-can-generate-human-voices-not-just-video.txt", "jsonld": "https://wpnews.pro/news/googles-new-ai-world-model-can-generate-human-voices-not-just-video.jsonld"}}