OpenAI has unveiled Sora 2, its upgraded artificial intelligence video generator, along with a new mobile application that allows users to produce and share AI-generated clips.
The app, called Sora, will initially be available in the United States and Canada through an invite-only rollout. It will feature a feed where users can create and remix videos, putting the company in direct competition with short-form content platforms.
Published in OpenAI’s blog post, the company said that Sora 2 is capable of generating video with synchronized audio and movements that follow natural physical laws.
Moreover, OpenAI said that the new system corrects previous flaws in sports and performance clips, where actions appeared unrealistic. A “cameo” feature will also allow users to insert their likeness into videos.
The company admitted that the app comes with risks similar to social media platforms, including overuse, exposure to harmful content, and misuse of likeness. Teen users will face default limits on video generation, while all users can control how their likeness is used.
Meanwhile, the launch comes as OpenAI faces criticism over how the first Sora was trained. Rights groups and industry analysts suspect that copyrighted material may have been used without permission.
No lawsuits have been filed directly against Sora, but the company is already dealing with separate copyright cases tied to its other models.
Analysts say that pairing Sora 2 with an app signals OpenAI’s attempt to control both the technology and its distribution, raising questions on how AI-generated video could affect existing platforms.
They also warned about the possible spread of misleading or manipulated clips as the line between real and synthetic video narrows.
OpenAI said Sora 2 would also be made available through an application programming interface in the coming months. On the other hand, the original Sora model will remain accessible to users.
