The video-generation AI model "Sora," developed by OpenAI, was briefly exposed to the public before its official release. Reports indicate the leak occurred during testing by visual artists.

According to the Financial Times (FT), the incident took place through a project hosted on the developer platform Hugging Face. An API connection inadvertently allowed external users to access and use Sora, a cutting-edge model developed earlier this year but withheld from public release due to unresolved copyright issues.

Currently, only a limited group of visual artists, designers, and filmmakers have access to Sora for testing purposes. They are tasked with providing feedback on potential risks, improvements, and impacts. However, some participants have criticized OpenAI for leveraging hundreds of artists to experiment with the model without proper compensation.

These artists, part of the so-called "Red Team" — a term for early-stage AI testers — allege that they were recruited without pay. Referring to themselves as "puppets for promotion," they accused OpenAI of engaging in "artwashing," a practice where corporations use art to enhance their public image or portray themselves as socially responsible.

As a result of the controversy, OpenAI has reportedly revoked access to Sora for participating artists.

The FT highlighted this incident as an example of the growing tension between AI developers and artists, who fear significant losses from such technologies. OpenAI’s leadership, including CEO Sam Altman, has previously tried to assuage concerns within the entertainment industry. Earlier this year, they showcased Sora to Hollywood studios while addressing fears about AI potentially displacing jobs in filmmaking.

Last month, over 11,000 cultural figures, including actors Julianne Moore and Kevin Bacon and Radiohead frontman Thom Yorke, signed an open letter declaring that generative AI poses a “serious and unjust threat” to the livelihoods of creators.

By Yonhap News