The new AI social network from ChatGPT: OpenAI has announced the launch of Sora 2 – the next generation of its AI video creation model – along with a unique app that functions as a closed social network. Unlike other platforms, all content appearing in the feed will be generated solely by artificial intelligence, even if the videos feature real people.
The app, currently available only on iPhone, will operate like a social network: Users can register, follow friends, and share content. However, all such content will be generated by the new model. One of the main features drawing attention is called “Cameo,” which allows users to integrate their own faces or their friends’ faces into AI-generated scenes.
Those who allow their face to be publicly used may also appear in other videos. For the first time, it will also be possible to generate videos with full sound, a feature that places Sora 2 alongside Google’s Veo 3, launched this summer. According to OpenAI, the new model significantly improves handling of physics laws, a previously problematic area in AI video generation systems.
The new engine can produce more realistic movement, objects that respond naturally, and more logical scenarios in a filmed environment. Anyone interested in joining will require an invitation. Each active user will receive four codes to share, gradually expanding access to the app.
Android users will only be able to log in via a browser, and they will also need an invitation code. OpenAI already recommends downloading the app from the App Store and registering with a ChatGPT account to receive a notification as soon as a code becomes available.
The bigger question is not only about creative possibilities but also about what the daily user experience will look like. OpenAI stated it will take a “conservative” approach regarding content moderation, in an attempt to prevent platform abuse. Blocks have been implemented to prevent the creation of pornographic or violent content. Additionally, every video exported from the system will feature a unique watermark to allow identification as AI-generated, even if uploaded to other platforms.
This marks OpenAI’s first step into the social networking field, a domain where companies like Meta already integrate AI-generated content on a massive scale. Meta has incorporated AI characters into Messenger, search, and its augmented reality glasses, effectively creating an entire feed filled with automated content. The difference is that Sora aims to offer a more human connection, where users themselves manage the content, while the tools enabling creation remain entirely artificial.
The first version of Sora was launched in 2024 but did not receive major updates. Since then, OpenAI integrated image generation capabilities into ChatGPT, sparking a global trend where millions of users created self-images in a Studio Ghibli style. The move was virally successful but raised numerous ethical and legal questions regarding copyrights and the use of real people’s likenesses.
Now, with the launch of Sora 2, OpenAI enters a new phase where it not only provides a creative tool but also offers a complete social platform entirely based on AI-generated content. The question that will follow the project in the coming months is whether users will embrace the platform as a new way to communicate and express themselves, or if it will become just another addition to the growing pool of social networks flooded with “junk content” originating from artificial intelligence.