The group appears to have leaked access to Sora, OpenAI’s video generator, protesting what it calls duplication and “art-washing” on OpenAI’s part.
On Tuesday, the group published a project on the AI dev site Hugging Face that appears to be connected to OpenAI’s Sora API, which is not yet publicly available. Using their authentication tokens – possibly from an early access program – the group created a virtual environment that allows users to produce videos with Sora.
Why I think it’s true – this uses the OpenAI Sora API endpoint to generate and download videos with hard-coded request headers and cookies from the Hugging Face local setting.
— Tibor Blaho (@btibor91) November 26, 2024
Using the group’s frontend, any user can produce 10 videos up to 1080p resolution. When TechCrunch tried it, the queue was quite long – but a few users on X were able to upload samples.
Try it here:
If it’s Sora, it looks like a modified version. It can create up to 1080 10-second clips.
Suggest repeating the space (if that works – my test doesn’t work!).
One example: pic.twitter.com/npphRJgyrd— Kol Tregaskes (@koltregaskes) November 26, 2024
So why did the group do this? They say OpenAI is pressuring Sora’s original testers, including red team players and smart partners, to tell a good story about Sora and failing to compensate them fairly for their work.
“Hundreds of artists provide unpaid work through bug testing, feedback and testing work [Sora early access] $150B value system [sic] company,” the group wrote in a previously attached post. “This early access program seems to be less about creative discourse and criticism, and more about PR and advertising.”
Confirmed: OpenAI Sora is indeed leaked pic.twitter.com/mAN1Z4vGsN
– Chubby♨️ (@kimmonismus) November 26, 2024
The group also claims that OpenAI is misrepresenting Sora’s potential by keeping early access users on a tight leash. All Sora output needs to be approved by OpenAI before it can be shared, they said, and a few creators in the program will be selected to have their Sora-created works evaluated.
“We are not against the use of AI technology as an artistic tool (if we were, we probably wouldn’t have been invited to this program),” he wrote. “What we don’t agree with is how this artist program was made and what the tool looks like before it is released to the public. We are sharing this with the world in hopes that OpenAI will become more open, artist-friendly and support the arts beyond PR stunts.”
We’ve reached out to Hugging Face and OpenAI for comment and will update this piece once we hear back.
Since the beginning of this year, Sora has faced technical challenges as competitors in the video production space work hard to overcome it. It doesn’t help, one of Sora’s main leads, Tim Brooks, left OpenAI for Google in early October.
In a recent Reddit AMA, OpenAI chief product officer Kevin Weil said Sora was caught up in “the need to perfect the model, get security/impersonation/other things right, and scale.” FYI, the original show, which aired in February, took more than 10 minutes to make a one-minute video clip.
The leaked Sora appears to be a faster, “turbo” variant, according to code obtained by X users.
Despite technical obstacles, OpenAI has released a significant partnership base for video production competitors in recent months. In September, Runway signed a deal with Lionsgate, the studio behind “John Wick,” to train a custom video model on Lionsgate’s catalog of movies. About a week later, Stability, which makes its own set of video production models, hired “Avatar” director James Cameron to its board.
OpenAI is said to be meeting with filmmakers and Hollywood studios earlier this year to showcase Sora; former CTO Mira Murati went to Cannes. But the company is yet to announce a partnership with a major production house.
