Sora 2: The First AI Social Network
The internet is on fire with the launch of Sora 2. However, the story isn't about the video model; it's about the Sora app.
Sora 2 is out. It isn’t just another AI video model; it’s OpenAI’s attempt to build an AI-powered TikTok. And that is kind of terrifying :)
The most interesting aspect of this launch is how OpenAI integrated marketing and virality into the product.
Let’s dive into some of the most interesting points, as we’re just over 48 hours from launch.
1. The “AI TikTok” Play
Sora 2 is comparable to Veo3. It creates very slick 8-second videos. What’s different is how OpenAI wrapped community features around the product to drive virality around the product.
As part of the onboarding process, it does two clever things: it scans your face to create an AI model (of you) and samples your voice. Previously you had to use a collection of tools to clone yourself and voice. Sora 2 does it instantly turns you into an AI creator ready to star in videos.
They call this their Cameo feature, making it easy to create videos with you as the main character, and more importantly, allowing you to insert other members of the app in your videos.
Those features move Sora 2 more towards a AI social network vs. just another AI video model:
- It creates viral moments among the community. You can easily create videos featuring other members of that community.
- When you include someone else’s AI model in your video, that person gets automatically tagged. It helps to create viral moments within the app and outside of the app (as most of them are getting traction on X right now).
- It solves the consistency problem Veo3 has across multiple clips. For example, I’m creating a 3-minute video with Veo3. It takes a lot of prompt work to keep characters consistent across those clips. Sora 2 solves this by having you create an AI model of yourself. The consistency is evident in every video that features Sam Altman, which is 99% of the content right now.
Here was the most viewed Sora video within a few hours, Sam Altman caught on CCTV stealing GPUs.
2. The Cameo Feature & Creator Economy Shift
OpenAI has just integrated with Stripe for payments. That’s a huge story. But, imagine how this fits with their vision for Sora 2 and the Cameo feature. It’s potentially a huge shift in how we think about the creator economy.
You don’t need to film yourself anymore; you can license your AI self. That means monetization without participation, and potentially a whole new market for “synthetic influencers”.
This is similar to the direction Meta is moving towards. It’s no surprise OpenAI poached some of Meta’s top engineering talent who likely worked on this app.
It’s not just creators who can license out their IP, but brands will be able to do this as well. Imagine having an army of creators incentivised to create short-form videos featuring elements of your brand. It makes user-generated content much easier to scale.
Here’s one of my favourite examples featuring Mario
3. The Video Quality Leap
Beyond the social mechanics, the model itself is a big step up from Sora 1: better lighting, physics, realism, and crucially, character consistency across clips.
Here is a great video that shows Sora 1 vs. Sora 2.
Sora 1 (Volley Ball):
Sora 2 (Volley Ball):
The difference is unbelievable. You just need to check out the entire launch video created via Sora 2. It’s a huge step improvement.
For marketers, this is the part that matters most. Even if you ignore the social app, the underlying model will flow into APIs. Brands will be able to produce higher-quality synthetic video ads and short-form video at scale.
Of course, we Europeans will eventually get our hands on it when the market has moved onto Veo4 and/or Sora 3 :(
Until Next Time,
Happy AI’fying
Kieran