OpenAI has announced the release of Sora 2, its flagship video and audio generation model, representing a significant leap from the original Sora introduced in February 2024. Described as the 'GPT-3.5 moment for video,' Sora 2 demonstrates advanced capabilities in physical accuracy, including complex gymnastics and realistic physics simulations that respect object permanence and dynamics. Unlike previous models that might 'morph' reality to fulfill a prompt, Sora 2 accurately models failure, such as a basketball rebounding off a backboard rather than teleporting into the hoop.

The release includes a new standalone iOS app, simply named 'Sora,' which features a social feed and a 'characters' function. This feature allows users to insert their own likeness into generated videos after a one-time verification process. To address concerns regarding digital wellbeing and safety, OpenAI is implementing natural language-controlled recommendation algorithms, default limits for teen users, and robust consent controls for personal likeness usage.

This launch signifies a major milestone in world simulation technology and the evolution of co-creative AI platforms.