OpenAI Gives Content Owners More Say in Sora, will Share Revenue

OpenAI is rolling out new features for its video-generation app Sora, so that owners of films, TV shows, or characters can control if and how their work is used—and even earn money when it is.

Mohammedia – OpenAI launched Sora as a tool that can generate short videos (up to 10 seconds) from text prompts—sometimes including copyrighted characters unless those owners opt out. The company now says it will let rights holders block uses of their characters or allow them under conditions, and will test a revenue-sharing model with those who agree to let users generate content with their characters.

Under the updated rules, if a studio or creator doesn’t want their character used, they can block it. If they allow it, they could receive a cut of whatever users generate when using their characters. OpenAI CEO Sam Altman admits they’ll have to experiment to find a fair way to compensate creators.

This matters because many critics have said AI tools like Sora risk misusing copyrighted material. By giving content owners more control, OpenAI is trying to strike a balance between letting users be creative and protecting creators’ rights.

Still, concerns won’t vanish overnight. The default in Sora is to allow use unless a creator opts out—which many see as putting the burden on rights holders rather than requiring explicit permission first. Also, enforcing those controls (detecting misuse, sharing revenue correctly) is harder than it sounds.

Furthermore, since Sora makes very realistic videos, there’s worry about misuse. Videos with violence, racism, or deepfake content have already appeared on public feeds of the app. This raises questions about moderation and whether controls will be strong enough to stop harmful content.

OpenAI also faces scrutiny over its training of Sora. Observers have found signs that the model can reproduce stylized logos or scenes that look similar to existing copyrighted works—suggesting it might have been trained on protected content. OpenAI says it used publicly available and licensed data, but hasn’t disclosed exactly which sources.

Whether the new control and monetization approach will earn creators’ trust depends on execution. If revenue splits are fair and blocking is effective, creators might see value. But if controls fail or compensation is weak, this could become another battleground in the AI vs IP wars.