OpenAI's Sora will be ready to churn out AI-generated videos for you this year

Earlier this year, ChatGPT creator OpenAI showed off its new text-to-video generative AI model called Sora. At the time, the AI tool was only available to red teamers. However, it seems the company is planning to publicly launch the tool pretty soon.

In an interview with The Wall Street Journal, OpenAI’s chief technology officer (CTO) Mira Murati spoke about the company’s plans for Sora. During that interview, Murati mentioned the AI tool would be available “this year” and that it could be ready in a few months.

The CTO went on to say that OpenAI wants to make it possible for users to be able to edit the videos produced by Sora. And down the line, the firm is also planning to eventually incorporate audio into those AI-generated videos.

It appears Sora is a costly AI model to run, as Murati told the publication it is “much more expensive” to power than other models. Despite that, OpenAI is reportedly aiming to make the tool “available at similar costs” to DALL-E — its text-to-image AI tool.

Like other generative AI tools, Sora needs to be trained on data for it to produce relevant content. When asked about what data was used to train Sora, the CTO seemed hesitant to provide an answer. “I’m not going to go into the details of the data that was used, but it was publicly available or licensed data,” Murati told the WSJ. While she did confirm that Sora uses content from Shutterstock — which OpenAI has a partnership with — she wasn’t sure if Sora also borrowed from YouTube, Facebook, and Instagram.

As for guardrails to help prevent the spread of disinformation, Murati says videos will have a watermark and it likely won’t be able to generate images of public figures. Whether this will be enough to deter disinformation remains to be seen, but watermarks have been broken before.