OpenAI has launched Sora 2, its latest text-to-video (and now audio) model, alongside a new iOS social app, Sora, that lets people create and remix short AI-generated clips in a TikTok-style feed.
What’s new in Sora 2
- More realism & physics: OpenAI says Sora 2 is more physically accurate and controllable than prior versions, better at keeping objects and scenes consistent over time. (OpenAI)
- Video with sound: The model generates synchronized dialogue and sound effects to match the scenes it creates.
- Availability: Sora 2 powers the new Sora app and is also being rolled out via limited invitations; OpenAI says broader access will expand over time.
OpenAI’s standalone Sora app features a vertical video feed and “Remix” tools for collaborative creations. Early access is limited to the U.S. and Canada, with invite codes circulating among testers. Several outlets describe it as OpenAI’s most direct foray into consumer social video to date.
OpenAI published a Sora 2 system card outlining risks such as misuse of personal likenesses and deceptive content. Initial deployment restricts uploads featuring photorealistic people and tightens moderation around content involving minors, with a staged rollout to monitor real-world use.
Sora 2 pushes AI video closer to production-quality output by coupling visuals with coherent audio and tighter scene control—capabilities that could accelerate advertising, entertainment prototyping, creator workflows, and social content. But the companion app also puts OpenAI in the social arena, raising fresh questions about consent, attribution, and platform safety.
Discover more from MultiMedia
Subscribe to get the latest posts sent to your email.
Leave a Reply