OpenAI has announced its new AI video generation model, Sora 2, which combines visual and audio elements in AI-created videos for the first time. The company also introduced the Sora iOS app, giving users the ability to generate and share short AI videos with friends through features that emphasize creativity over prolonged screen time.
According to OpenAI's website, Sora 2 improves significantly on the original model launched earlier this year by offering more accurate physical simulations that eliminate unrealistic effects such as teleporting objects. It supports diverse visual styles ranging from anime to photorealistic rendering. For example, missed basketball shots in AI-generated clips now feature realistic ball bounces off the backboard rather than teleporting to the hoop. Official demos presented at the launch displayed complex, dynamic movements across sports like gymnastics, beach volleyball, skateboarding, and paddleboarding, showcasing the system's enhanced ability to capture nuanced motion details.
Sora app focuses on creative engagement with social interaction
Alongside the model, OpenAI debuted the Sora app, designed to provide users with a personalized, creative video experience instead of conventional social media's attention-driven feeds. Users can create videos up to 10 seconds long from text or images that can be remixed and further developed.
One notable feature is "Cameo," which enables multiple users to appear in the same AI-generated scene remotely, creating a shared social experience without physical presence. To participate, users must record video and audio of themselves, complete identity verification, and allow permission for others to use their likeness in these AI scenes. OpenAI ensures that no AI-generated image or video of a real person appears without their explicit consent and identity confirmation. Users maintain full control over their data and can revoke or delete any associated content at any time.
CEO Sam Altman acknowledged the risks posed by the platform, including addiction and potential bullying. He pledged that OpenAI would closely monitor Sora's usage and potentially modify or discontinue the service if it negatively impacts users' well-being.
Sora currently operates by invitation only on iOS devices in the US and Canada, with plans to expand access to additional regions. The Sora 2 API is expected to launch in the coming weeks, allowing third-party developers to integrate the technology into professional editing tools. An Android version of the app is in development, promising new features such as storyboard control for creators.
Copyright concerns spark industry tensions
The launch of Sora has reignited debates over copyright and AI-generated media. The Wall Street Journal reported that OpenAI has informed film studios and agencies that copyrighted materials will be included in Sora content by default unless copyright holders explicitly opt out—a continuation of policies previously applied to AI image generation. Reuters cited insiders confirming that Walt Disney has opted out, blocking its copyrighted content from appearing on the platform.
Industry analysts, including those at Morgan Stanley, observe that OpenAI's Sora challenges established social and digital content giants like Meta, Google, and TikTok. The move broadens OpenAI's reach into entertainment and media sectors, aiming to grow its user base beyond traditional AI technology users.
As Sora expands, the balance between innovation, user safety, and intellectual property rights will remain a crucial factor shaping its adoption and future development.
Article edited by Jerry Chen
