In late September, OpenAI launched the second version of its revolutionary platform “Sora2”, an artificial intelligence system that will radically transform the world of video creation thanks to its exceptional capabilities.
After starting as a silent, video-only pilot in February 2024, the project has evolved into a full-fledged consumer product, now delivering synchronized sound and hyper-realism that redefines visual media creation.
The platform has enjoyed unprecedented popularity, as the number of downloads of “Sora2” exceeded one million in less than five days, surpassing ChatGPT in terms of the speed of initial spread. The application also topped the Apple App Store list in the markets where it was launched.
However, the app is still only available by invitation to iOS device users in the US and Canada, allowing OpenAI to manage pressure on servers and monitor potential cases of abuse.
At its core, Sora2 relies on an advanced artificial intelligence technology known as Diffusion Transformers, which processes so-called “space-time patches,” which are small data units that combine spatial and temporal dimensions. This approach allows the system to have a deep understanding of motion while keeping the appearance of objects consistent across shots.
One of the most notable improvements in the new version is the sound that is completely synchronized with the movement of the lips in dialogues, in addition to advanced physical simulation that produces amazing realistic interactions that could not be achieved before.
The platform also supports multiple types of input, going beyond text to include moving still images and editing video clips with high precision. Output capabilities have been developed to reach 1080p resolution with a duration exceeding one minute, making the technology suitable for both personal and professional use.
To ensure transparency, all videos produced include C2PA digital signatures that clearly identify the content as being generated by artificial intelligence, in a move aimed at combating disinformation using advanced technical tools.
“Sora2” works as a standalone application on iOS and as a web platform as well, as the application provides an interface similar to TikTok for creating and sharing artificial intelligence-generated content.
The Cameos feature is one of the most attractive features, as it enables users to insert their faces and voices into the created videos, which contributed to the widespread spread of the platform, but at the same time raised ethical questions about privacy, consent, and the use of personal photos.
The company has faced some criticism regarding its copyright policies. Following feedback from rights holders regarding the previous “opt-out” policy, OpenAI has adopted a new model based on voluntary subscription and revenue sharing, providing an innovative monetization path for IP holders.
The company plans to expand the platform globally with the launch of an Android version, providing APIs to developers, making Sora 2 the foundation for a whole new environment of AI-driven media applications.
With this rapid progress, OpenAI continues to develop more robust watermarking and content monitoring systems, to mitigate legitimate concerns about misinformation or synthetic media abuse.
As “Sora2” passes its limited experimental phase, it is expected to make a wide impact in the fields of entertainment, marketing and communication, as this pioneering technology represents the beginning of a new era in digital media production, where the boundaries of human creativity and artificial intelligence merge, opening unprecedented horizons for creators around the world, and forever changing our relationship with the arts of visual narrative.
