OpenAI Launches Sora 2 AI Model with TikTok-Style Video App
OpenAI has introduced Sora 2, its upgraded video generation model, nearly a year after the launch of the first version. Unlike its predecessor, this release adds native audio generation, placing it alongside advanced models like Google’s Veo 3. The company described the launch as a “GPT-3.5 moment for video,” emphasizing its ability to manage complex movements while respecting realistic physics.
A demo highlighted the technology’s strength, showing a gymnast performing flips on a balance beam with lifelike motion. Moreover, the company stressed that Sora 2 avoids object distortions, reducing hallucinations and producing videos that remain consistent with prompts. The model is optimized for cinematic, realistic, and anime styles, and it now supports instructions across multiple shots without breaking scene continuity.
Launch of the Sora App
To showcase the model, OpenAI also launched a TikTok-inspired app called Sora, currently available on iOS through invite-only access in the US and Canada. The app offers a short-video feed where all clips are AI-generated using Sora 2. Users can upload a video of themselves, and the system can place them into any AI-generated setting while maintaining strong character consistency.
The app includes interactive features like likes, comments, shares, and a Cameo function. Cameo allows users to take characters from other videos and generate new clips with fresh prompts, effectively creating customizable deepfakes of friends. Videos can also be generated directly through the Sora website after invite approval.
Safety and Identity Verification
While Sora’s features encourage creativity, safeguards have been included. Before generating personal videos, users must verify their identity to prevent unauthorized use of celebrity or random content. However, the Cameo option bypasses some of these protections, raising questions about potential misuse. OpenAI noted that the tool feels like a natural progression in communication—from text and emojis to voice notes and now personalized AI video.
Currently, Sora 2 video generation is free for all invited users, with plans to expand availability later. By blending powerful AI video tools with a social media format, the company aims to redefine how people create and interact with content online.
