OpenAI’s Sora App Lets Friends Swap AI Video Cameos

OpenAI just launched an app that lets you create videos of yourself and your friends doing things you’ve never actually done. The new Sora app comes with a feature called cameos that puts you into any scene you can imagine, and you can give your friends permission to use your face in their videos too.

The app arrived on Tuesday alongside Sora 2, which is OpenAI’s newest video creation tool. Instead of just being able to make random videos, you can now star in them yourself. Want to see yourself skateboarding down a mountain even though you’ve never touched a skateboard? Or maybe show your friend dunking a basketball when they can barely reach the hoop? The cameo feature makes all this possible.

Here’s how it works. You record a short video of yourself talking and moving around just once. This recording captures what you look like and how you sound. The system uses this to create videos where you appear to do all sorts of things. Once you’ve done this setup, you can drop yourself into any scene the app can generate.

The really interesting part is that you can share your cameo with friends. If you trust someone enough, you can give them permission to put your face in videos they create. This means your best friend could make a funny video showing both of you on an adventure together, even if you’re in different cities. It opens up lots of creative possibilities for making content with people you know.

OpenAI designed the app to work like a social network, similar to TikTok or Instagram Reels. You can scroll through a feed of videos that other people have made and shared. The app shows you content based on what you’ve liked before, where you’re located, and even your past conversations with ChatGPT if you use that. You can turn off the ChatGPT connection if you don’t want your chat history affecting what videos you see.

The new Sora 2 model is much better at creating realistic videos than the first version. Earlier video makers would sometimes do strange things like make a basketball magically teleport into the hoop if a player missed a shot. The new version actually follows the rules of how things work in real life.

OpenAI showed off examples of beach volleyball games, skateboard tricks, gymnastics routines, and people jumping into pools. All these videos look surprisingly real, with movements and physics that make sense. The improvement over older versions is pretty obvious when you watch the clips.

Right now, the app is only available in the United States and Canada, and you need an invitation to use it. OpenAI plans to expand to more countries soon, but they’re starting small to work out any problems before opening it up to everyone. If you’re already paying for ChatGPT Pro, you can try out the Sora 2 video maker without needing an invitation, though the social app part still requires an invite.

The app is completely free to use at launch. OpenAI wants people to explore what’s possible without worrying about costs. The only time they might charge you is if lots of people are trying to make videos at once and the system gets overwhelmed. In that case, you might need to pay to generate extra videos during busy times.

Parents should know that the app comes with some controls for keeping kids safe. You can limit how long they scroll through videos, turn off the system that picks what videos they see, and control who can send them messages. These tools only work well if parents actually know how to use them, though.

There are some obvious concerns about this technology. Giving someone permission to use your face in videos requires a lot of trust. Even if you trust a friend now, things can change. Someone could create embarrassing or harmful videos of you if they wanted to. You can take back permission at any time, but the damage might already be done if they’ve shared something before you revoke access.

The system tries to prevent people from creating videos of others without permission. You can’t just type in someone’s name and make a video of them. They have to actually go through the verification process and give explicit permission for their likeness to be used. This is meant to stop people from making fake videos of celebrities or people they know without consent.

Despite these safety measures, there’s still risk involved. Videos showing people doing things they never did can spread quickly online. Once something is out there, it’s hard to take back. OpenAI will need to watch carefully for abuse and be ready to step in when problems happen.

The company says they built this social app because they think cameos are the best way to experience what Sora 2 can do. Creating videos with yourself and friends is more engaging than just making random scenes. It gives the technology a personal touch that might make people want to use it more often.

This launch puts OpenAI in direct competition with other social video apps. TikTok dominates short video content, and Instagram Reels has carved out a big space too. Even Meta just launched a similar video feed called Vibes in their AI app last week. Everyone seems to be rushing to capture attention in this space.

The difference with Sora is that everything is created by artificial intelligence instead of people filming real videos. Whether users will embrace this or prefer authentic recorded content remains to be seen. Some people might love the creative freedom of making impossible videos. Others might find it less appealing than watching real moments captured on camera.

For now, OpenAI is focused on getting people excited about the possibilities. The free access and invite-only approach creates buzz and exclusivity. As more people get invitations and start playing with cameos, we’ll see how the app catches on and whether friends really want to star in each other’s generated videos.