OpenAI has launched Sora, an innovative text-to-video AI model capable of producing videos up to 60 seconds long based on text prompts.

Sora employs diffusion modeling and leverages DALL·E 3's recaptioning technique to create accurate captions for its visual training data.

The AI model can animate still images, augment existing video content, and reconstruct missing video frames.

OpenAI's CEO, Sam Altman, acknowledges Sora's current limitations, such as directional confusion and challenges in visual consistency.

The release of Sora has stirred a mixed reaction among creative professionals, sparking discussions on its impact on creativity and the potential for spreading misinformation.

Sora's debut highlights the rapid progress in AI video generation while also prompting societal and ethical considerations for its use and regulation.

OpenAI has published a technical report titled 'Video generation models as world simulators' to detail Sora's technology and capabilities.