Only a week and a half after OpenAI launched the new Sora AI image generator app, the company (which also runs ChatGPT) is flip-flopping on its most controversial feature: that creators who didn’t want their work to appear in user-generated Sora videos had to explicitly tell OpenAI that they opted out.
It may be only nine days later by the time I write this, but it’s about time.
Videos by VICE
a rapid about-face
“We are going to make two changes soon (and many more to come),” wrote OpenAI CEO Sam Altman on his personal blog on October 3. “First, we will give rights holders more granular control over generation of characters, similar to the opt-in model for likeness but with additional controls.”
The Sora app, which launched on September 30, initially operated on an opt-out model that I criticized as soon as it launched. Opt-out requires a person to first be aware that their content is fair game for people on Sora to use, and then for them to also navigate to the Sora website and click a button, slider, or menu to remove their content from the platform.
If a person or company is unaware of Sora or doesn’t fully comprehend what it can do with their content, too bad. The consequences can be shocking and unexpected. It’s how you end up with a Sora-generated video of SpongeBob SquarePants spouting hateful rhetoric.
Opt-in works the opposite way. By default, everybody’s content is not fair game for Sora users to take, shape, and mold like pixelated clay. But if a person or a company wants to allow people to use their content in making Sora videos, they can navigate to the Sora website and toggle a setting that allows it. It’s a fairer way to organize the app.
By the time of Altman’s post, Sora was already generating chaos from videos of fake Middle East bombings, fake ballot box stuffings, Nazi SpongeBob, and the like.
“Second,” Altman continued, “we are going to have to somehow make money for video generation. People are generating much more than we expected per user, and a lot of videos are being generated for very small audiences.”
“We are going to try sharing some of this revenue with rightsholders who want their characters generated by users. The exact model will take some trial and error to figure out, but we plan to start very soon. Our hope is that the new kind of engagement is even more valuable than the revenue share, but of course we want both to be valuable.”
