OpenAI announces video generation AI 'Sora 2,' and the simultaneously released iOS social app 'Sora' can generate videos from your own and your friends' faces with Sora 2



OpenAI announced Sora 2 , a new cutting-edge video and audio generation model, on September 30, 2025. Additionally, Sora , a social video sharing app that utilizes Sora 2, has been released for free on iOS.

Sora 2 is here | OpenAI

https://openai.com/index/sora-2/

Sora 2 System Card | OpenAI
https://openai.com/index/sora-2-system-card/

The Sora feed philosophy | OpenAI
https://openai.com/index/sora-feed-philosophy/

Sora 2 is here | OpenAI
https://openai.com/index/sora-2/

Sora 2 - Sam Altman
https://blog.samaltman.com/sora-2

Sora 2 is built on the foundation of Sora and offers new features that were difficult to achieve with traditional video models, such as more accurate physics, realistic depictions, synchronized audio, enhanced operability, and an expanded range of styles.

The following movie is a video actually generated by Sora 2.

This is Sora 2 - YouTube


One of the biggest advancements in Sora 2 is its ability to simulate the physical world. While previous generations of models would deform objects or ignore the laws of physics to fulfill text prompts, Sora 2 has evolved to follow more realistic physics.

You can also incorporate real-world elements directly into Sora 2. For example, simply load a video of a teammate and Sora will accurately recreate their appearance and voice and incorporate them into the environment it generates. OpenAI claims this feature is so general that it can be applied to any person, animal, or object.

In addition, OpenAI says Sora 2 has improved the fidelity with which it follows user commands. It boasts 'enhanced steerability' and can accurately respond to complex commands across multiple shots while maintaining a consistent worldview. It also has expanded capabilities for a wide range of styles, including realistic live-action, cinematic, and anime.

In addition, Sora 2 can simultaneously generate audio as well as video, allowing realistic ambience, dialogue, and sound effects to be synchronized with the generated video with high fidelity.

OpenAI has released the following movie that demonstrates the synchronization of video and audio generated by Sora 2.

Sound on for Sora 2 - YouTube


Regarding safety, OpenAI commented, 'In developing Sora 2, we have built more robust safety measures by leveraging what we learned from the previous generation of Sora.' Specifically, the company has introduced moderation classifiers that comprehensively monitor input prompts and output video and audio, as well as stricter safeguards to protect minors.

Additionally, content generated by Sora 2 is marked with C2PA metadata, an industry-standard proof of provenance, and downloaded videos display a visible, moving watermark indicating that they are AI-generated.

Sora 2 is available through the iOS app 'Sora,' which OpenAI released on the same day. The app is currently available in the United States and Canada, with plans to expand to other countries in the future.

Sora by OpenAI on the App Store
https://apps.apple.com/us/app/sora-by-openai/id6744034028



Sora is not just a video generation app, but a social app where users can share the videos they create. The in-app feed is designed to encourage users to create their own content, rather than simply consume it, and is intended to provide new ideas and inspiration by viewing other users' work.

The app's feed prioritizes content from people you follow and interact with, and users can also adjust the algorithm to suit their mood and control the content they see.



There's also a 'Cameo' feature that lets you include friends or yourself in Sora-generated video environments. This Cameo feature gives you 'full control over who can use your likeness,' meaning you can revoke permission or delete videos that include your likeness at any time.

X has posted a video generated by Sora showing OpenAI CEO Sam Altman being caught by a security guard trying to take a GPU out of a store without paying.



Additionally, users under the age of 18 will be limited to a certain number of videos they can watch per day, and the Cameo feature will be subject to strict restrictions. The app also features a parental control feature that allows parents to manage their children's usage settings via ChatGPT. Additionally, videos downloaded from the app will have a watermark.

In a blog post, Altman wrote, 'We're a little worried, too. Social media has had both positive and negative effects on the world. We know how addictive these services can be, and we can imagine the many ways they could be used for bullying. It's easy to imagine the worst-case scenario: AI-generated videos eventually drag us all into a cluttered, real-world optimized feed. Our team has put a lot of care and thought into how to create a compelling product that doesn't fall into that trap.'



Users who register within the Sora app will receive a push notification when access to Sora 2 begins, and will then be able to access Sora 2 on the web app , Sora.com . While the service is free, OpenAI says, 'We have initially set sufficient limitations so users can freely explore its capabilities, but computational limitations still apply.'

In addition, users who subscribe to ChatGPT Pro, a monthly subscription plan costing $200 (approximately 30,000 yen), can also use the 'higher quality Sora 2 Pro model,' which is available experimentally on Sora.com. OpenAI also announced that it plans to release an API for Sora 2.

Related Posts:

in Video,   Software,   Smartphone, Posted by log1i_yk