Fake videos are about to look less fake, for better or worse. On Tuesday, OpenAI announced the release of Sora 2, the latest version of its flagship model for audio and video generation. And, as previously reported, the launch of the model is accompanied by a new social app designed to allow people to share their AI-generated videos, creating an endless scroll of uncanny content that will almost surely further fry the brains of people.
In an announcement video that the company claims was completely generated by Sora 2, a fabricated version of OpenAI CEO Sam Altman called the model “the most powerful imagination engine ever built.” The focus of the update appears to be what OpenAI calls “world simulation,” attempting to accurately recreate the physics of the real world. The company put a strong emphasis on highlighting videos of people moving in realistic ways. The company admits it’s still imperfect, but claims Sora 2 is “better about obeying the laws of physics compared to prior systems.” OpenAI also claims the model is much better at following intricate instructions and can now produce multiple different shots based on a prompt.
Sound on. pic.twitter.com/QHDxq6ubGt
— OpenAI (@OpenAI) September 30, 2025
Then there’s the Sora app. As rumored and reported on by Wired, the Sora app is a feed entirely of videos created via OpenAI’s video generation model. The app features a vertical scroll to move through videos, which are served up based on a recommendation algorithm. Users will be able to insert themselves into a video through a feature called “cameo,” which requires users to record a video of themselves to verify their identity and, in OpenAI’s words, “capture your likeness.” Insert ominous needle drop here. Other users will also be able to use your likeness in videos.
While your likeness can be used by others, OpenAI insists that the user is in control. “Only you decide who can use your cameo, and you can revoke access or remove any video that includes it at any time. Videos containing cameos of you, including drafts created by other people, are viewable by you at any time,” it said.
OpenAI did make a point to insist that it is rolling out the Sora social app “responsibly.” Users will be able to set their own feed by telling the model what they want to see, and the company claims it’s “not optimizing for time spent in feed, and we explicitly designed the app to maximize creation, not consumption.” Creation is also time spent in-app, of course, but whatever. The company also claims that teenage users will be subjected to strict limits, with a cap on how many videos they can view per day and restrictions on how their likeness can be used.
The company also said it doesn’t currently have a plan to monetize the app… except for the plan it has to monetize the app. “Transparently, our only current plan is to eventually give users the option to pay some amount to generate an extra video if there’s too much demand relative to available compute. As the app evolves, we will openly communicate any changes in our approach here, while continuing to keep user wellbeing as our main goal.” Basically, there will eventually be a limit on how much you can generate unless you pay for more. And it’s not like it’s particularly difficult to see ads working their way into the feed, but the company isn’t specifically saying that will come (though it also doesn’t explicitly rule it out).
The Sora app, which is invite-only for now, can be downloaded on iOS by users in the US and Canada, with plans to roll out beyond North America soon. It’ll be free to start. As for Sora 2 itself, it’s available for ChatGPT Pro users (that’s the $200 per month tier) for now.