The makers of ChatGPT are now also creating video from text specifications

Chatbot makers ChatGPT have filled the gap in their offering with software that can create videos from text specifications. The AI ​​model called Sora will initially be made available to selected creative people, Sam Altman, head of OpenAI, wrote Thursday on Internet platform X (formerly Twitter). Experts should also explore potential security risks before using the software on a large scale.

Videos created by Sora can be up to one minute long. On the program's website, OpenAI posted several examples along with the descriptions they were based on. One shows a woman walking down the street. The video was created entirely by artificial intelligence, with the text specifying that the woman should wear a leather jacket and a red dress and that the street should be reminiscent of Tokyo and have a lot of neon signs that are also reflected in the puddles. Other videos show, among other things, mammoths walking in the snow and a California town during the Gold Rush.

Many other companies have already developed software that can create videos from text. OpenAI admits that Sora still has weaknesses: the model sometimes makes mistakes when implementing the rules of physics. It could also happen, for example, that someone takes a bite of a cookie in a video — and the cookie still looks whole later.

AI technology that generates animated images from text templates could change video production over time. At the same time, there are major concerns that fake videos could be created on a large scale, and can be difficult to distinguish from real recordings. So technology developers are working on ways to incorporate unique identification features like watermarks into videos. Sora's videos must also show that they were created by artificial intelligence.

ChatGPT has been generating hype around AI for just over a year. Just like AI chatbots, image and video creation software is also trained on vast amounts of information.