OpenAI has announced the release of Sora 2, a video generation model, and the new Sora app . Positioned as a social creation platform, the new app supports short videos generated from text (currently capped at approximately 10 seconds), Remix creations, and Cameo authoring. Users can verify and control the use and visibility of their own likenesses. Officials claim Sora 2 offers significant improvements over its predecessor in terms of physical consistency, controllability, and audio-visual synchronization, and plans to gradually expand creative capabilities and duration. The app is currently available by invitation only in the United States and Canada, with plans for iOS and Android releases in the works.
Regarding security and governance, OpenAI announced its launch principles and mitigation measures: The creation of content related to public figures and sensitive/illegal content will be restricted, explicit consent will be required for inclusion, and periodic checks will be implemented for usability and emotional impact. Violating content will be intercepted through a multi-layered approach using models, rules, and manual review. Officials emphasized that the system will be adjusted based on actual feedback, aiming to ensure ease of use and efficiency while avoiding an addictive "information sludge."
Frequently Asked Questions
Q: What is the relationship between Sora and Sora 2?
A: Sora 2 is a video/audio generation model; Sora App is a creation and sharing application based on this model.
Q: Who can use it now?
A: Invited users are available in the United States and Canada, and the number of places will be gradually expanded; each user can invite a small number of friends.
Q: How does Cameo prevent abuse?
A: You need to verify and grant permission to use it yourself. The generation of portraits without consent is prohibited and public figures are restricted.
Q: What is the video length and function?
A: The current upper limit is about 10 seconds, supporting text generation, remix and authorized entry; subsequent functions will be expanded according to the official roadmap.
Q: Are there any safety and compliance measures?
A: It includes copyright and portrait consent, sensitive content blocking, emotion/addiction risk monitoring and manual review, etc. The details will be updated with the version.
