OpenAI stated that it will make two major changes to Sora, reflecting feedback from users and rights holders. First, the company will implement a feature allowing rights holders to exercise finer control over generated characters. While this mechanism is similar to similarity settings for facial photos, it differs in that the scope of use can be specified more concretely.
According to OpenAI, while some rights holders are hopeful about this new "interactive fan fiction" format, many others wish to decide for themselves how their characters are used, including the option for complete exclusion. OpenAI intends to apply the same standards to all rights holders, allowing each to choose the optimal approach. The company also mentioned the deep connection between Japanese content and its users, expressing high regard for their creative achievements.
Next, OpenAI addressed the monetization of video generation. The company noted that the volume of generation per user has exceeded expectations, and many videos are being created that attract few viewers. OpenAI plans to introduce a model—through a process of trial and error—where a portion of the revenue from characters generated by users is shared with the rights holders. While the company aims for this new engagement itself to hold more value than the revenue, it indicated a commitment to prioritizing both.
OpenAI stated that rapid changes, similar to the early stages of ChatGPT, may occur. While there may be both good decisions and mistakes, the company's policy is to receive feedback, quickly correct errors, and iterate with Sora, consistently applying these learnings to other products.
Source: Sora Update #1 (HN 137pt, 177 comments) (HN Search (backfill), 2025-10-04)