OpenAI unveils Sora text-to-video AI model
The company behind ChatGPT has unveiled a new artificial intelligence tool that can create videos based on text prompts.
It seems like every day there’s a new advancement in the artificial intelligence space. While we aren’t quite at the stage of Asimov robots, OpenAI’s latest development certainly blurs the line between film and video created by humans and something conjured up by a computer.
OpenAI announced its new AI model, Sora, via X (formerly Twitter) on February 15, 2024. In the multi-post thread, the company goes into some detail about what this new artificial intelligence tool can do and offers up several examples of it in action.
Introducing Sora, our text-to-video model.
— OpenAI (@OpenAI) February 15, 2024
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/7j2JN27M3W
Prompt: “Beautiful, snowy… pic.twitter.com/ruTEWn87vf
The above video is based on a text prompt describing a snowy Tokyo street bustling with people. The prompt also includes camera movement and details like Sakura petals and snowflakes. The thread is full of these videos using vastly different prompts from an instructional cooking session and woolly mammoths charging through snow to a sci-fi shot on 35mm film and a Pixar-like animated monster.
https://t.co/rmk9zI0oqO pic.twitter.com/WanFKOzdIw
— Sam Altman (@sama) February 15, 2024
While this is impressive, it’s not without its growing pains and classic AI foibles. For example, one video of a stylish woman walking down a Tokyo street shows her legs switching sides while the above video of the cooking session sees the woman conjuring a spoon out of thin air. However, the video does show off a human with the traditional number of fingers.
Prompt: “A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. she wears a black leather jacket, a long red dress, and black boots, and carries a black purse. she wears sunglasses and red lipstick. she walks confidently and casually.… pic.twitter.com/cjIdgYFaWq
— OpenAI (@OpenAI) February 15, 2024
There have been text-to-video AI tools around for a while now, but OpenAI’s appears to be another step along this strange path. Though the company is working with “domain experts in areas like misinformation, hateful content, and bias” no one can really say how these tools will be used by the general public and whether bad actors will be able to exploit them. The Biden Administration has even gone so far as to sign an executive order for new AI safety standards with one of the points being that developers must share safety test results with the U.S. government.
But there is pushback from some companies regarding regulation of artificial intelligence. Del Complex is one such company that plans to create floating computers that operate as sovereign nation states to avoid AI regulation.
Though OpenAI is using what it calls “red teamers” to test its Sora functionality and safety, the company’s tools have previously been used by hacking groups. Just recently, Microsoft has taken steps to ban hacking groups from Russian, Iran, China, and North Korea that it discovered were making use of OpenAI and other AI tools.
Artificial intelligence, and specifically AI-led tools, is still in the infancy stage of its life, despite how advanced the systems and modelling are becoming. Be sure to take a look at our OpenAI page for the latest on this company, especially as Microsoft continues to push more out of it, like a web search service to compete with Google.
-
Sam Chandler posted a new article, OpenAI unveils Sora text-to-video AI model