OpenAI has significantly expanded the horizon of generative AI with the launch of its new Sora 2 text-to-video generation model. The newly launched model succeeds the original Sora model, which was introduced in 2024. Alongside the new video generation model, the company has also released a companion social app simply called Sora.
Sora 2: A leap in realism
According to the company, Sora 2 can accomplish tasks that were exceptionally difficult—and in some instances, outright impossible—for prior video generation models. While previous models often had to morph objects and deform reality to successfully execute a text prompt, OpenAI claims that Sora 2 is better at obeying the laws of physics, which is an extremely important capability for any useful "world simulator".
Enhanced control and sound generation
The company also states that Sora 2 represents a major leap forward in controllability, able to follow intricate instructions spanning multiple shots while accurately maintaining the world state. Furthermore, it is capable of creating sophisticated background soundscapes, speech, and sound effects with a high degree of realism.
Injecting real-world elements
Sora 2 can also directly inject elements of the real world. The model can accurately insert real-world objects and voices into any Sora-generated environment with an accurate portrayal of their appearance and voice.
Still evolving
OpenAI has acknowledged that the model is far from perfect and makes plenty of mistakes. However, the company views the launch as validation that further scaling up neural networks on video data will bring them closer to simulating reality.
The social app 'Sora'
OpenAI has also launched a new social iOS app named "Sora," which is powered by Sora 2. Inside the app, users can create and remix each other’s generations and discover new videos in a customizable feed. The app also allows users to feature themselves or their friends via cameos.
Launching responsibly
The company has prioritised user safety by providing tools and options for users to control what they see on their feed. Using OpenAI's existing large language models, the company developed a new class of recommender algorithms that can be instructed through natural language. Furthermore, it has built-in mechanisms to periodically poll users on their well-being and proactively give them the option to adjust their feed.
By default, the app shows users content heavily biased toward people they follow or interact with and prioritizes videos that the model thinks users are most likely to use as inspiration for their own creations.
Protecting the well-being of teens is a key concern. The company has put in default limits on how many generations teens can see per day in the feed and is rolling out stricter permissions on cameos for this group.
Availability
The Sora iOS app is available to download now. Users can sign up in-app for a push notification when access opens for their account. The company is starting the initial rollout in the U.S. and Canada with the intent to quickly expand to additional countries. Sora 2 will initially be available for free, with generous limits to allow people to freely explore its capabilities.
ALSO READ: Who is Zoho Arattai founder Sridhar Vembu? Lives in village despite Rs 50,000 crore net worth
