OpenAI has officially launched Sora 2, its latest and most advanced video and audio generation model, marking a significant leap forward in AI-powered content creation. Building upon the foundation of its predecessor, Sora 2 introduces a suite of new capabilities designed to offer greater realism, improved physics simulation, synchronized audio, and enhanced steerability for users. This new model aims to empower storytellers and creators with a more powerful toolkit, while also pushing the boundaries of AI's ability to simulate the complexities of the physical world.

The enhanced features of Sora 2 include more accurate physics, sharper visual realism, and the ability to generate synchronized audio, making the generated videos more immersive and believable. The model demonstrates a high degree of fidelity in following user prompts, allowing for the creation of videos that are both imaginative and grounded in real-world dynamics. This advancement is seen as a crucial step towards developing AI models that can deeply understand and interact with the physical environment.

In response to concerns and potential risks associated with advanced AI generation, OpenAI is implementing a phased and iterative approach to safety. Sora 2's initial rollout includes limited invitations, restrictions on the use of photorealistic person image uploads and all video uploads, and stringent safeguards for content involving minors. The company is actively working with internal red teamers to identify potential risks, such as nonconsensual use of likeness or misleading content, and is developing corresponding mitigations.

OpenAI is also addressing copyright concerns by offering copyright holders more "granular control" over how their work is used by Sora 2. Following initial reports of the model generating videos featuring popular copyrighted characters, CEO Sam Altman announced that copyright holders will be able to opt out of having their intellectual property recreated. This system aims to be similar to the opt-in model for generating AI "cameos" of individuals, allowing creators to specify how their characters can be used, or if they can be used at all.

Sora 2 is being made available through sora.com, a new standalone iOS app, and is planned for future API integration. OpenAI is prioritizing user safety and responsible deployment, with ongoing efforts to refine the system based on user feedback and evolving understanding of AI capabilities and risks. The company aims to balance maximizing creative potential with ensuring safety and ethical use of the technology.