OpenAI now also generates video from text presets

Software from the makers of the chatbot ChatGPT can generate videos from text specifications. The AI ​​model called Sora will initially be made available to selected creatives, OpenAI boss Sam Altman wrote on the online platform X on Thursday.

Experts must also investigate potential security risks before the program can be widely used, as Altman wrote on the online platform X (formerly Twitter). Videos created by Sora can be up to one minute long.

On the software website, OpenAI published several examples, along with the description on which they were based. One of them shows a woman walking down a street. The video is generated 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 lots of neon signs that are also reflected in puddles. Other videos include mammoths walking in the snow and a California town during the gold rush.

Several 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 physics rules. For example, it could also happen that someone in the video takes a bite of a cookie – and that the cookie still looks intact later.

Be careful with fake videos

AI technology that generates moving images from text templates could transform video production over time. At the same time, there is great concern that fake videos could be created on a large scale and be difficult to distinguish from real recordings. The developers of the technology are therefore working on ways to include unique identifying features, such as watermarks, in the videos. Sora videos must also show that they are created by AI.

ChatGPT fueled the hype around artificial intelligence just over a year ago. Like such AI chatbots, the photo and video generation software is also trained with vast amounts of information. (sda/dpa)