On the system card for the Sora 2 AI model (which powers the Sora app), OpenAI says it was trained on things found on the internet:
“Sora 2 was trained on diverse datasets, including information that is publicly available on the internet, information that we partner with third parties to access, and information that our users or human trainers and researchers provide or generate.”
This seems like an invitation for a big copyright lawsuit, along the lines of the one Disney, Dreamworks, and NBCUniversal recently filed against AI image generator Midjourney.
But OpenAI is trying to flip the responsibility of protecting copyrighted material to the intellectual property owners themselves. According to WSJ, OpenAI is allowing copyrighted material in Sora by default, unless copyright holders opt-out of the service.
The courts will have to decide if this novel approach to intellectual copyright law works, but government regulators may not be that big of a problem, as Sam Altman has made sure OpenAI is in the good graces of the Trump administration. If OpenAI has to pay up to copyright holders after a lawsuit, what’s a few billion dollars here or there when you are raising so much capital?
