In February, OpenAI first announced Sora, its new AI video creation app. It allows people to make videos of up to 60 seconds by just typing in some text prompts. The examples that OpenAI posted to show off videos created via Sora impressed many people with its realistic visuals, camera movements, and more.
So far, OpenAI has only allowed a few invited users to try out Sora, in part because the company wants to make it safe to use. However, in a new interview with The Wall Street Journal, OpenAI"s chief technology officer, Mira Murati, stated the current plan is to launch OpenAI to the general public sometime later in 2024. In fact, she said it "could be a few months" before Sora"s public launch.
One of the concerns about Sora, and in fact for all text prompt-based AI tools, is where the data is coming from for their large language models. In the case of Sora, Murati said that it does use content from Shutterstock that OpenAI has licensed. However, she declined to reveal any more data sources, saying they come from "publicly available or licensed data."
Right now, Sora can only create silent videos, and those clips cannot be edited during the creation process. Murati did say OpenAI is working on adding audio to Sora"s clips, along with editing tools. Right now there"s no word on how much OpenAI will charge for using Sora to the public. According to Murati, OpenAI would like to price Sora similarly to how it prices its DALL-E 3 AI art creator. However, Mira added that Sora is "much more expensive" for OpenAI in terms of power and support.
Of course, OpenAI"s biggest financial and development partner is Microsoft, which has incorporated its Chat-GPT and DALL-E models in its Copilot services. Just after Sora was announced, someone asked Mikhail Parakhin, Microsoft"s head of Advertising and Web Services on X (formerly Twitter), if Sora would be added to Copilot as well. Parakhin replied, "Eventually, but it will take time."