The new social app from OpenAI created for generating AI short-form videos has taken the world by storm. Just over a week since Sora’s release, it seems a new update is already on the way.
What’s happened? Sam Altman, the CEO of OpenAI, the same company that founded ChatGPT, has announced new updates for the newly-launched Sora in his blog.
- First, the update will bring more control for rightsholders, such as Disney and other companies with famous IP, to choose how their characters or brand can (or can’t) be used in AI-generated content.
- Second, there will be revenue for rightsholders who allow their characters to be generated by users. Altman states Sora has to have a revenue stream, and while the actual model on how this will work is yet to be determined, this could be a good platform for rightsholders.
- Altman also cautions that the platform will experience a “very high rate of change” as the company try things, make failures, gathers input and course-corrects.
Why is it important? The first change – granular controls for rightsholders – is going to dramatically change what the generative AI is able to create. If a brand like Disney or Studio Ghibli decides it doesn’t want its IP included in the model, it won’t be able to generate videos using their characters.
Why should I care? If you’re already creating videos in Sora, or are keen to check it out, what you’re able to generate is about to shift.
- We’ll have to wait and see if rightsholders are open to allowing Sora to use their IP in its AI-generated videos, or whether we’ll see big names refuse to opt-in.
- If rightsholders are reluctant to allow their IP to be used in the videos Sora creates, the appeal of the platform could diminish.
- However, if OpenAI is able to convince brands to get on board with an enticing revenue model, we could see Altman’s vision of “this new kind of ‘interactive fan fiction’” explode in popularity.
What’s next? Altman confirms these updates are on the horizon and coming very soon, as the team continues to iterate on various methods and transfer lessons across all their products.
- The revenue-sharing model remains unspecified and will change based on feedback and experimentation.
