OpenAI Made Video Creation Effortless—Here’s What Happened Next

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Sam Altman is here. Sam Altman is there. Sam Altman is whispering in your ear.

He’s also stealing GPUs from a Target, barbecuing prawns at a company retreat, and co-starring in a sitcom with your best friend.

I, for one, would like a day—just one—without Sam Altman in my feed. But that looks highly unlikely in the near future because of Sora 2.

OpenAI’s new flagship video-and-audio model launched a couple of weeks ago, and now everyone’s starring in AI-generated videos of themselves—along with, apparently, Sam Altman.

Sora 2 is OpenAI’s first new video model since the release of Sora Turbo almost a year ago. On the same day it released Sora 2, OpenAI also released an invitation-only social iOS app, also called Sora, that’s powered by the new model. If you grant the app permission to use a recording of your face and voice, your friends can drop you into their videos as a “cameo”—and you can return the favor. Another popular feature, “remix,” lets users build on each other’s clips: You can take a video your friend generated and put your own spin on it.

Think of it as Instagram for your AI self. Instead of sun-drenched selfies and pictures of brunch, your feed is full of videos of you and your friends in all kinds of wacky invented situations—like having a horse ride them on the moon or approaching Altman—yes, him again—at OpenAI’s DevDay.

You can access Sora at sora.com, via the Sora iPhone app, and, per OpenAI’s announcement at DevDay, in the API. Read on for a look at what people on the Every team think about Sora, along with the most interesting takes from across the internet.

What everyone at Every is thinking

Sora stands out because it’s the first video model that can tell a story—and invite you in

“Sora feels like the first truly entertainment/storytelling generative model, not just visually, but narratively… What sets it apart isn’t just the output, it’s how the input works—unlike other models where the burden of creativity sits on the prompt, Sora takes that load off you. You can throw in a vague prompt, like ‘Make me a funny video,’ and the model somehow figures out the story beats, characters, and flow. The underlying prompt system feels invisible, like there’s a hidden narrative engine that interprets intent rather than instruction. That also makes it a black box; you never really know what’s happening behind the scenes, which can feel frustrating but it’s also part of the magic.”—Lucas Crespo, creative lead

The only generative video model that’s full of your friends