A samoyed dog and a golden retriever are playfully walking around a futuristic neon city at night.


OpenAI, which entered the mainstream last year thanks to the popularity of ChatGPT, is bringing its AI technology to video.

The company on Thursday presented Sora, its new generative AI model. Sora works similarly to OpenAI’s AI imaging tool, DALL-E. The user enters a desired scene and Sora will return a high definition video. Sora can also generate videos inspired by still images and expand existing videos or fill in missing footage.

Video may be the next frontier for generative AI now that chatbots and image generators have made their way into the consumer and business worlds. While the creative possibilities will excite AI enthusiasts, the new technology raises serious misinformation concerns as major political elections around the world approach. The number of AI-generated deep fakes created increased 900% year over year, according to data from Clarity, a machine learning firm.

With Sora, OpenAI aims to compete with artificial intelligence tools for video generation from companies like Meta and Googlewhich Lumiere announced in January. Similar AI tools are available from other startups, such as Stability AI, which has a product called Stable Video Diffusion. Amazon also launched Create with Alexa, a model that specializes in prompt-based, short-form animated content for kids.

Sora is currently limited to generating videos that are a minute or less in length. OpenAI powered by Microsofthas made multimodality—combining text, image, and video generation—a goal in its efforts to offer a broader set of AI models.

“The world is multimodal,” OpenAI COO Brad Lightcap told CNBC in November. “If you think about the way we as humans process the world and engage with the world, see things, hear things, say things – the world is much bigger than text. So for us, it’s always felt incomplete for text and code to be the single modalities, the single interfaces that we could have to how powerful these models are and what they can do.”

Until now, Sora has only been available to a small group of safety testers, or “red teams,” who test the model for vulnerabilities in areas such as misinformation and bias. The company has not released public demonstrations beyond 10 sample clips available on its website, and said accompanying technical documentation would be released later Thursday.

OpenAI also said it is building a “discovery classifier” that can identify Sora-generated video clips, and that it plans to include certain metadata in its output that should help identify AI-generated content. This is the same type of metadata that Meta wants to use to identify AI-generated images this election year.

Sora is a diffuse AI model that, like ChatGPT, uses the Transformer architecture presented by Google researchers in a 2017 paper.

“Sora serves as the foundation for models that can understand and simulate the real world,” OpenAI wrote in its announcement.

WATCHING: OpenAI is on its way to a ‘true technological breakthrough’

Don’t miss these stories from CNBC PRO: