OpenAI's Sora video generator was briefly leaked on Hugging Face
OpenAI announced its text-to-video generative model, Sora, in February this year. OpenAI said during the unveiling that Sora could produce videos up to a minute in length and handle intricate scenes with multiple characters and motion.
While OpenAI hasn't made any official announcement regarding a public release, an anonymous group of artists reportedly shared the model online on the Hugging Face machine learning repository.
Techcrunch reported the model was online for roughly three hours before OpenAI revoked access. Not everyone could beat the queues to generate videos, but some users uploaded examples similar to what OpenAI showed.
The group reportedly leaked the model because they believed they were being forced to "tell the world that Sora is a useful tool for artists."
The group says that OpenAI shut early access to all artists after the leak. The open letter says, "While hundreds contribute for free, a select few will be chosen through a competition to have their Sora-created films screened — offering minimal compensation which pales in comparison to the substantial PR and marketing value OpenAI receives."
The letter says the group wants to "denormalize" artists being used as free PR and unpaid R&D for "billion dollar brands." The group also encouraged people to sign a petition that asks big brands to "provide a path to true artist expression, with fair compensation to the artists."
Source(s)
