TikTok clips, animation and all things filmed – video is a ubiquitous medium with huge influence, but the knowledge and equipment needed to put ideas on a screen can be expensive and hard to come by.

Thanks to artificial intelligence (AI), these barriers are being lowered.

Artificial intelligence (AI) industry leader OpenAI announced their newest project “Sora” today. The text-to-video model interprets a written prompt and generates actual footage.

For the user, it functions the same way as ChatGPT, a prompt-to-text model, and DALL·E, a prompt-to-image model, both OpenAI projects.

Currently, Sora can generate 60 seconds of video per prompt. The current version can generate a video using a still photo, extend and fill frames from real videos and create videos based on written prompts.

The AI is a diffusion model. It takes the prompt, finds images related to it, breaks those images down and then the AI rebuilds the images in the way it “wants” them to look.

In the earlier stages of diffusion AI development, like DALL·E, images could be recreated, but signage and text were distorted. Improvements were made in later versions of the model, DALL·E 2 and 3, and the information pool they use feeds Sora too.

While the software brings with it opportunities and makes video work accessible to all, it brings increased risk as well.

Deep fake imagery is causing increasing concern. Of the 95,000 deep fake videos surveyed for a study last year, 98% were deep fake pornography. 99% of those explicit videos were of women. The cases of high profile streamers and celebrities, including Taylor Swift, have, however, shone considerable light on this issue recently.

While legislation around the world is slowly playing catch-up with the technology, Sora’s development team have written-in some preventative measures, relating to its potential use for the generation of violent and sexually explicit content.

“Despite extensive research and testing,” Sora’s home page says, “we cannot predict all of the beneficial ways people will use our technology, or all the ways people will abuse it.”

Sora is in the early stages of development and is not yet open for public access.

Story by Anisha Satya.