OpenAI, the company behind ChatGPT, has launched Sora 2, its latest AI-driven video model, along with a social networking app where all content is produced by artificial intelligence, a Tuesday announcement stated.
Sora 2 is the next generation of OpenAI's AI video model, paired with a new app that functions as a closed social network. Unlike other platforms, every post in the feed is created entirely by AI, even when it features real people.
Currently available only on iPhones, the app functions similarly to a typical social platform, allowing users to register, follow friends, and share content. However, all content is generated by the new AI model.
One standout feature is Cameo, which enables users to insert their own face or a friend's into AI-generated scenes. Additionally, users who choose to make their faces publicly available might appear in others' videos. For the first time, videos on the platform will feature full sound, bringing Sora 2 on par with Google’s Veo 3, released earlier this summer.
What does Sora 2 do?
According to OpenAI, Sora 2 significantly improves how AI handles the laws of physics—something previous systems struggled with. The updated model produces smoother motion, more realistic interactions with objects, and more logically sound scenarios in filmed environments.
Access to the platform is currently invitation-only. Each active user receives four invitation codes to share; OpenAI will gradually expand its availability.
Android users can join through a web browser, but will also need an invitation code. OpenAI encourages users to download the app from the App Store and register with a ChatGPT account to be alerted when an invitation code is available.
Beyond the creative features, OpenAI emphasizes the importance of user experience. The company has promised a "conservative" approach to moderation, designed to prevent the misuse of the platform. This includes built-in safeguards against the creation of explicit or violent content. Additionally, every exported video will feature a unique watermark, making it identifiable as AI-generated—even if uploaded elsewhere.
This marks OpenAI's first foray into social media, a space already populated by companies like Meta, which uses AI to generate large amounts of content across its platforms. Meta has incorporated AI-generated characters into Messenger, search tools, and its augmented reality glasses, creating feeds filled with synthetic content. However, Sora sets itself apart by aiming for a more personal experience, with users guiding the content—even if the underlying tools remain entirely artificial.
The initial version of Sora launched in 2024 but did not feature major updates. Later, OpenAI integrated image-generation capabilities into ChatGPT, triggering a global trend where millions of users created Studio Ghibli-style self-portraits. While this trend went viral, it also raised critical questions about the ethics of using real people's likenesses without consent.
With Sora 2, OpenAI is moving beyond offering creative tools to providing a fully AI-generated social platform. The big question in the months ahead will be whether users embrace this as a new form of communication and self-expression, or if the platform becomes just another feed filled with AI-generated "junk content."