Introducing Sora 2: OpenAI’s Next-Gen Video and Audio Generator

OpenAI has unveiled Sora 2, a cutting-edge audio and video generator designed to create more realistic and physics-aware content. Accompanying this release is a new social app, Sora, which allows users to create and share videos using a TikTok-style algorithmic feed.

Sora 2 demonstrates significant improvements in video realism, addressing past issues where video models distorted reality to meet text prompts. The model’s enhanced physics engine ensures that actions within the videos, such as a basketball shot, behave in accordance with real-world physics.

The Sora app introduces an innovative ‘cameos’ feature, enabling users to insert themselves into Sora-generated scenes. Users must first upload a one-time video-and-audio recording to verify their identity, after which they can share these personalized videos with friends.

Currently available in the U.S. and Canada, the app is invite-only, with plans to expand globally. ChatGPT Pro users, however, can access the Sora 2 Pro model without an invite. The app offers parental controls powered by ChatGPT, providing options to manage content and interactions.

Initially free, Sora plans to monetize by charging for additional video generation during high demand. OpenAI emphasizes the importance of user safety and gives users control over their likeness, though the potential for misuse remains a concern.

As the platform grows, OpenAI aims to deliver a secure and engaging experience, leveraging user activity and preferences to curate personalized content while ensuring privacy and safety.