OpenAI Unveils Technology for High-Quality Video Creation

OpenAI, the company that created ChatGPT, has developed an AI model called Sora that creates video clips of up to 60 seconds from detailed text instructions. The clips are of high quality, a sign that technology has advanced a lot in the last year. “Sora” is not yet available to the general public as negative implications are being looked into.
This AI model will initially be available to a small number of experts to test how the technology can be used for malicious purposes. Numerous negative scenarios will be analyzed to determine how serious situations can be avoided.

OpenAI has released several short clips showcasing what this model named Sora can do. You can give it a variety of commands, from “show me woolly mammoths running through the snow” to “create a video of wild animals competing in a bicycle race”.

OpenAI also has a program called DALL-E, which generates images based on user-given instructions. DALL-E was announced in January 2021, DALL-E 2 in April 2022, and DALL-E in October 2023.

The clips on Sora are not perfect; sometimes the movements are odd or “stilted”, but the technology has great potential, advancing significantly in the past year. The clips can be up to 60 seconds long, whereas the first AI-generated clips could be much shorter.

The company is aware of the technology’s enormous potential for deception, especially as the phenomenon of “deepfake” has gained momentum and become increasingly risky, particularly in the context of elections in many countries. In the announcement regarding the “Sora” technology, OpenAI has a “Safety” section.

OpenAI states that they are also working on technologies to analyze metadata to quickly detect when a video has been constructed using AI technologies.

Furthermore, text instructions requesting the creation of videos with violent or sexually explicit content, as well as requests involving celebrity names, will not be accepted.

Comments (0)
Add Comment