OpenAI Unveils Sora 2: Text-to-Video-and-Audio Model with Parental Controls for ChatGPT
OpenAI has launched Sora 2, a new text-to-video-and-audio model, along with an invite-only iOS app for social creation and consent-controlled 'cameos'. The model focuses on physical plausibility and synchronized dialogue. It's initially free but compute-constrained. Meanwhile, OpenAI has introduced parental controls for ChatGPT.
Sora 2, OpenAI's latest model, generates native, time-aligned audio along with video. It maintains state across shots, offering materially better world modeling than its predecessor. The model blocks generations featuring real people, except when users have opted in via the 'cameo' feature. All outputs carry C2PA metadata and a visible watermark.
The Sora iOS app, now available by invite, lets users record short videos to verify identity and capture likeness. Users control who can use their likeness and can revoke or delete videos. ChatGPT Pro users gain access to an experimental Sora 2 Pro tier with API access planned after the consumer rollout. Existing Sora 1 Turbo content remains accessible in user libraries.
OpenAI's Sora 2 brings enhanced text-to-video-and-audio capabilities, focusing on physical plausibility and multi-shot controllability. The accompanying iOS app offers consent-controlled 'cameos' for inserting verified likenesses. Alongside this, OpenAI has introduced parental controls for ChatGPT, allowing parents to manage their teens' usage.
