OpenAI Unveils Sora 2: Text-to-Video-Audio Model & Social Creation iOS App
OpenAI has launched Sora 2, a new text-to-video-and-audio model, along with a social creation iOS app for users in the US and Canada. The app, initially invite-only, focuses on physical plausibility and multi-shot controllability.
Sora 2, the successor to the popular Sora 1 Turbo, offers improved world modeling and maintains state across shots. It generates native, time-aligned audio, including speech, ambient sounds, and effects. The new Sora iOS app allows users to create, remix, and insert verified 'cameos' of themselves into generated scenes, with controls over who can use their likeness and revoke or delete videos.
OpenAI has integrated parental controls via ChatGPT, enabling parents to opt teens into a non-personalized feed, manage DM permissions, and control continuous scroll. Initially, the app is free under compute-constrained caps, with access opening by invite. OpenAI is restricting certain features at launch, such as image uploads featuring photorealistic people and all video uploads. Text-to-video generations of public figures are also blocked, except via the cameo feature. Existing Sora 1 Turbo content remains available in user libraries.
The Sora app, developed by OpenAI and launched in September 2025, carries C2PA metadata and a visible moving watermark on downloads, with internal detection tools for origin assessment. ChatGPT Pro users gain access to an experimental Sora 2 Pro tier, with API access planned post-consumer rollout. OpenAI plans to expand the app's availability to other regions gradually.
