OpenAI’s Sora appears to have been leaked.

In protest of duplicity and “art-washing” on OpenAI’s part, a group appears to have leaked access to OpenAI’s video generator, Sora.

On Tuesday, the group posted a project on AI development platform Hugging Face that appears to be connected to OpenAI’s Sora API, which has not yet been released. Using authentication tokens supposedly from the early access program, the group created a frontend that allowed users to create videos with Sora.

The group’s frontend allows any user to create 10-second videos at up to 1080p resolution. When I tried TechCrunch, the queue was quite long, but several users on X uploaded samples.

So why did the group do this? They claim that OpenAI is pressuring Sora’s early testers, including the red team and creative partners, to develop a positive story about Sora and failing to compensate them fairly for their work.

“Hundreds of artists are providing unpaid labor through bug testing, feedback, and experimental work on the (Sora Early Access) program for the $150 billion company,” the group wrote in a post attached to the frontend. “This early access program appears to be more focused on PR and advertising than creative expression and criticism.”

The group also claims that OpenAI is tying up early access users and misleading them about Sora’s capabilities. All Sora output must be approved by OpenAI before being shared, and only a small number of creators will be selected to screen Sora-created works in the early access program, they say.

“We are not opposed to using AI technology as an artistic tool. “If that were the case, we probably wouldn’t be invited to this program,” he wrote. “What we disagree with is how this Artist Program has been rolled out and how the tools are being structured ahead of public launch. We’re sharing this with the world in the hope 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 story when we hear back.

Since its debut earlier this year, Sora has experienced technical difficulties as its rivals in the video generation space have worked hard to overtake it. Not helping matters, one of Sora’s co-leaders, Tim Brooks, left OpenAI for Google in early October.

In an October Reddit AMA, Kevin Weil, OpenAI’s chief product officer, said Sora was hampered by “the need to perfect the model, get safety/impersonation/etc right, and scale compute.” The original system, Per The Information, released last February, took more than 10 minutes of processing time to create a one-minute video.

According to code discovered by user X, the leaked Sora appears to be a faster “turbo” variant of the Sora.

In addition to technology-related hurdles, OpenAI has ceded valuable partnership ground to video-generation challengers in recent months. In September, Runway signed a deal with Lionsgate, the studio behind “John Wick,” to train custom video models for Lionsgate’s catalog of films. About a week later, Stability, which is developing its own set of video creation models, brought “Avatar” director James Cameron to its board.

OpenAI reportedly held meetings with filmmakers and Hollywood studios to demonstrate Sora earlier this year. Former CTO Mira Murati attended the Cannes Film Festival. However, the company is yet to announce any collaboration with any major producer.