OpenAI surprises with the launch of the AI short video application, going live simultaneously with the new Sora 2.

OpenAI officially released Sora 2 today. This next-generation audio-visual generation model is not only more realistic and physically accurate, but it also enables synchronized dialogue and sound effects, with higher controllability. Along with the newly launched Sora App, users can create, remake, and interact with generated videos, even transforming themselves into the main character of the video.

Significant breakthrough in film generation

The Sora series was first launched in February 2024, showcasing the capabilities of video generation models in aspects such as object continuity and scene continuity. Building on this, Sora 2 further enhances the ability to handle complex situations that previous models struggled to generate correctly, such as Olympic-level gymnastics movements, backflips on a balance beam, and even completing a three-and-a-half jump while balancing a cat on the head during figure skating.

Sora 2 provides a more realistic physics simulation.

The previous video models often featured “exploiting loopholes” that violated physical laws. For example, after a basketball player misses a shot, the ball might instantaneously teleport into the hoop. Sora 2 allows the ball to bounce off the backboard, aligning more closely with real-world physics. This ability to represent “failure” rather than just “success” is regarded as an important milestone in the development of world simulators.

At the same time, Sora 2 significantly enhances controllability, maintaining a consistent world state across scenes and showcasing cinematic, animated-style realism; in addition, it can generate detailed sound effects and background soundscapes, making the audiovisual experience more complete.

Sora App and Cameo Features

OpenAI has simultaneously released the new Sora application, which is not only a video creation tool but also integrates social features. Through the Cameo function, users can record a short video verification once, allowing them to accurately embed themselves or friends into Sora scenes while retaining voice characteristics.

Internal testing shows that this feature generates more interaction among employees. The official believes this is a brand new generation of social interaction methods following text messages, emojis, and voice messages.

Safety and Health Design

OpenAI states that the design of the Sora App aims for “creation first” rather than “time-consuming viewing,” and introduces a recommendation system that can be adjusted through natural language. To avoid addiction and excessive use among teenagers, the platform has a daily viewing limit, parental control features, and a dedicated human review team to handle bullying or abuse.

In addition, users have full control over their “Cameo” and can revoke others' usage rights or delete related videos at any time, ensuring that their personal likeness and voice are not misused.

This article announces that OpenAI has surprisingly launched an AI short video application, which will be launched simultaneously with the brand new Sora 2, first appearing in Chain News ABMedia.