Last year OpenAI launched Sora and one year later OpenAI has introduced Sora 2 which is built directly from Sora. Along with the model, OpenAI has linked the release of Sora App which is a social media platform. This is an ambitious step taken where they are aiming to blend Generative AI with social networking. This announcement actually follows after months of speculation about OpenAI moving towards consumer facing platforms.
Sora 2: A More Realistic and Physically Accurate Model
Here’s the thing- OpenAI revealed in a blog post that prior video models were overoptimistic. They were so focused on the prompt that they would morph objects and deform reality. Which made the videos very unrealistic and not aesthetic. With the learnings from Sora, OpenAI is aiming to change the same drawback with Sora 2. Sora 2 is better at following laws of physics, which is going to make the videos more realistic, Techrunch reported. So now when a person playing basketball misses a shot, it will rebound off the backboard.
One of the cool features that the model brings is that it is great in high-complexing scenarios. Be it gymnastic routines, skateboard tricks, and cannonball jumps from a diving board.
The Sora App: Designed for Socializing
As mentioned above along with Sora 2, OpenAI is releasing an IOS app called Sora. Here’s what you can go with the App- you can generate, share videos just like how you do on Instagram and TikTok. But the defining feature which has the spotlight is Cameos. Cameos is going to let users upload their videos and generate AI clips and themselves and their friends.
To use Cameos, users have to first upload a video of themselves with an audio to verify their likeness. And once that’s done, they can start creating AI generated clips. Most important is the permissions part, users can revoke their cameo permission at anytime they want. So, users retain control always. Further, now that we are at the topic of safety let’s talk about other safety controls. OpenAI has actually integrated several features from ChatGPT to Sora to mitigate potential risks. Like for example, there’s parental control where parents can manage feed personalization and limit scrolling by setting a timer. Additionally, the app also asks users to periodically reflect on their experience and even adjust recommendations.
Monetization and Availability
At launch, the app is going to be free. And as per OpenAI, its plan for monetization is only on the days of high demand and that’s just to offset computing costs. This may change as usage scales, but right now the plan is to not engage itself as an engagement-revenue model.
The app is initially going to be out in first USA and Canada, with plans of expansion in the future. Currently access remains invite-only. While on the other hand ChatGPT Pro Subscribers can test out the Sora 2 model directly on sora.com.
The launch of the Sora App also raises questions about responsibility. Because here’s the truth- Non-consensual AI videos have been creating a lot of problems since the very beginning. With Sora and Cameo integration OpenAI needs to have strong moderation policies. The company says that right now one of the measures that they have taken is that they scaling human detectors. This is alongside automated filters to detect any potential harmful content. These factors are going to determine if Sora can be a trusted platform.
Sora vs. Sora 2: What’s New
| Feature | Sora (2024) | Sora 2 (2025) |
|---|---|---|
| Physics & Realism | Basic object permanence, errors common (e.g., teleporting objects) | Improved world modeling, realistic rebounds, buoyancy, and failures |
| Audio | Limited, basic soundscapes | Full synchronization: speech, effects, background audio |
| Controllability | Struggles with multi-shot continuity | Maintains world state across complex prompts |
| Visual Styles | Simple cinematic and animation | Advanced realism, anime, cinematic-grade visuals |
| Social Integration | None | Cameos, remixing, TikTok-style feed in the Sora app |
| Availability | API and research access only | iOS app (U.S. & Canada), invite-only feed, Pro access on sora.com |
Why This Matters
With Sora 2, OpenAI is not only planning on competing with other AI models, but also with social media platforms like Instagram and TikTok. This is because of the similar kind of feed it’s aiming to provide. And the ambitious move of integrating the Cameo feature suggests that they are ready to experiment generative AI with social media platforms.
And if Cameo gets widely adopted, Sora could shift how people look at video creation. The success of Sora is going to depend on factors such as if it’s engaging, something unique, and socially meaningful.
