Well, that was fast. One day after Wired reported that OpenAI was preparing to release a new AI social video app, the company has revealed it to the wider world. It's called the Sora app, and it's powered by OpenAI's new Sora 2 video model, allowing it to generate AI-made clips of nearly anything. As expected, the app's signature "cameo" feature allows people to add your likeness to videos they generate.
Cameos are likely to be controversial, even if OpenAI is giving users a lot of control over whether someone can replicate their likeness in clips Sora generates. When you first start using the app, you can allow your friends (and even strangers) permission to generate images of you. Whenever someone uses your likeness in a video Sora will designate you as the "co-owner" of that clip, allowing you to later delete it or prevent others from further modifying the video with subsequent generations. The latter plays into Sora's "Remix" feature, which allows users to jump on trending videos to offer their own take on them. Sora 2 can generate sound alongside video, a first for OpenAI's model.
Separately of the above restrictions, Sora can't generate videos of public figures — unless they upload their likeness to the app and grant their friends or everyone permission to use it in their creations — and the software will refuse to make pornographic content.
Right now, Sora is only available on iOS, with no word yet on when it might arrive on Android, and you'll need an invite from the company. However, those lucky few who can join are able to invite four friends to download the software, much like the early days of say, Bluesky or Clubhouse (lol). OpenAI is only making Sora available to people in the US and Canada (sorry, everyone else).