Earlier this year, OpenAI announced Sora, a text-to-video AI model that showed off detailed scenes and complex camera movements from relatively simple prompts. It’s been radio silent since then, but the company recently granted free early access to the tool to artists for testing. However, OpenAI suspended access after a group of about 20 of them leaked their access to Sora, complaining that it was acting as a “PR puppet,” the Washington Post reported.
“We received access to Sora with the promise of being an early tester, red team, and creative partner. But in return, we received access to Sora with the promise of being an early tester, red team, and creative partner. “We believe that people are being seduced by ‘artwash’,” the group said. I wrote this on the AI art repository site “Hugging Face.”
He pointed out that OpenAI’s valuation recently reached $150 billion, and noted that artists are providing free testing and feedback. The group also pointed out that all content generated by Sora must be approved by OpenAI and is “focused on PR and advertising rather than creative expression.” The organization said it has since released the tool for anyone to play and hopes OpenAI will “go beyond PR to support the arts.”
In response, OpenAI shut down initial Sora access after just three hours while investigating the situation. “Hundreds of artists in alpha have helped shape the development of Sora and prioritize new features and safety measures,” OpenAI spokesperson Nico Felix said in a message to the Post. “Participation is voluntary. , there is no obligation to provide feedback or use the tools.” Another artist participating in the program, Andre Allen Anjos, similarly protested, saying that the protest artists’ stance did not reflect the views of most artists participating in the program.
Although Sora is not yet widely used, the tool has been scrutinized for training materials. Mira Murati, OpenAI’s chief technology officer, said in March that she did not know whether Sora obtained training data from YouTube or other video platforms. The following month, YouTube’s CEO specifically warned OpenAI that training models on videos was against its terms of service.
