OpenAI is actively learning and iterating on Sora based on user and rightsholder feedback. Soon, they will implement more granular controls for rightsholders regarding character generation, similar to likeness opt-in but with expanded options. This allows rightsholders to specify how their characters can be used, including the ability to prohibit usage entirely. The company anticipates a variety of approaches from users for this new form of "interactive fan fiction." They acknowledge that some unintended generations may occur during this iterative process. Additionally, a revenue model for video generation is necessary due to higher-than-expected usage and small audience sizes. OpenAI plans to share some of this revenue with rightsholders whose characters are generated by users. While the exact revenue-sharing model requires experimentation, it will launch shortly. The hope is that this new engagement proves more valuable than the revenue share, though both are being pursued. Users should expect rapid changes, akin to the early days of ChatGPT, with quick fixes for missteps. These iterative improvements will be tested on Sora before being applied across other OpenAI products.