A new AI model has demonstrated the ability to generate highly realistic and coherent video sequences from simple text prompts. The system, developed by researchers at a leading technology institute, uses a novel diffusion architecture that builds upon recent advancements in image generation. Initial demonstrations show the model creating short clips of animals, landscapes, and urban scenes with convincing motion and temporal consistency. The researchers acknowledge current limitations, including video length and resolution constraints, but highlight the potential applications in film, gaming, and simulation. Ethical considerations regarding deepfakes and misinformation are noted as areas requiring further safeguards and policy development. Read the full article at: https://technologyreview.com/ai-video-generation-breakthrough