Sora, the AI video program from OpenAI, has dominated many of the news headlines in the AI world in recent weeks.
Although it has so far only been made available to select artists, designers and filmmakers, spectacular videos created by Sora have already made the rounds on social networks.
Sound should also come – but when is still unclear
Soon there could be many more. In one Interview with the Wall Street Journal OpenAI Chief Technology Officer Mira Murati shared that Sora will be available to everyone “later this year.” Murati also revealed that OpenAI plans to include sound to make the video sequences even more realistic.
Editorial recommendations
When exactly this should be possible, she left it open with a very vague “at some point”. Another goal is to give users the ability to edit the content of the videos created by Sora.
Evasive answer to a sensitive question
Murati became much more reserved when she was asked about sensitive topics surrounding artificial intelligence – for example when asked which data OpenAI used to train Sora. “I won’t go into the details of the data used, but it was publicly available or licensed data,” she said.
Were videos from YouTube, Facebook and Instagram also used? Murati’s answer was that she wasn’t sure. At least she confirmed that Sora uses content from the stock provider Shutterstock, with which OpenAI has a partnership.
That’s how much Sora should cost
How much Sora will cost exactly is not yet known. According to Murati, the tool will be offered “at a similar cost” to Dall-E, OpenAI’s AI text-to-image model. As with Dall-E, it should not be possible for Sora to create imitations of public figures.
There are also plans to watermark videos created with Sora to make it clear that they are not real video material. However, it remains questionable whether this is enough to minimize the risk of confusion: The Verge has concluded that watermarking is not the perfect solution.