OpenAI’s Sora 2 Video Model Generates Audio and Has Its Own App

A person holding a megaphone is giving directions on a film set, surrounded by bright lights and filming equipment in a dimly lit environment.
An AI-generated Sam Altman grabbed his AI megaphone to announce Sora 2 yesterday.

OpenAI has launched Sora 2, its latest AI video generator that will further undermine people’s belief in what is real and what isn’t.

CEO Sam Altman says that “creativity could be about to go through a Cambrian explosion, and along with it, the quality of art and entertainment can drastically increase.”

The company livestreamed the announcement yesterday, following it up with a sizzle reel showing what Sora 2 is capable of. See below.

Unlike the first Sora model, the new one generates sound that can be synchronized with the video. It also has its own app called Sora, where users can create shareable 10-second video clips based on prompts or photos. Inside the app, users can create, remix other people’s videos, and discover new videos in a customizable Sora feed.

AI videos are known for their mistakes, and while OpenAI says the physics of Sora 2 are still “imperfect,” it is an improvement on past models. To emphasize the point, OpenAI shared a generated video of a person doing gymnastics, a subject AI has struggled with in the past.

There is also a new “Cameo” feature in Sora 2, which allows users to insert themselves or others into videos. However, anyone wanting to appear in an AI video will be required to verify their identity by a one-time video and audio recording of themselves.

“The team worked very hard on character consistency. The cameo feature is something we have really enjoyed during testing, and is to many of us a surprisingly compelling new way to connect,” Altman writes on X.

“Video models are getting very good, very quickly,” OpenAI says in a blog post. “General-purpose world simulators and robotic agents will fundamentally reshape society and accelerate the arc of human progress. Sora 2 represents significant progress towards that goal.”

Earlier this week, The Wall Street Journal reported OpenAI informed Hollywood studios that unless they explicitly opt out, their copyrighted content may appear in Sora output. It comes after Disney and Warner Bros. launched a lawsuit against the AI image generator Midjourney for allowing users to generate their copyrighted characters.

Right now, the Sora app is invite-only and exclusive to iOS. That will likely change in the coming weeks.

Discussion