Facing backlash, OpenAI reverses course: rightsholders will decide whether their copyrighted characters can be generated by the Sora 2 video app, with revenue sharing for those who opt in.
What a wild week. Just three days after launching Sora 2 with a brazen policy that let users create videos featuring copyrighted characters unless rightsholders explicitly opted out, OpenAI has slammed the brakes. Late Friday, the company announced it will move to an opt-in model requiring permission before copyrighted characters can appear in Sora 2 videos.
BACKGROUND—I tested OpenAI’s new Sora 2 video app. Within minutes I was generating branded cartoon clips. Plus, check out my interview with Puck’s Matt Belloni: Is Sora 2 the Entertainment Industry’s Next “Lazy Sunday” Moment?
In a blog post, CEO Sam Altman promised “more granular control” for rightsholders and floated revenue-sharing for those who opt in. The post framed the reversal as collaboration, not capitulation. No apology, no acknowledgment of overreach, just “learning quickly from how people are using Sora and taking feedback from users, rightsholders, and other interested groups.”
“We are hearing from a lot of rightsholders who are very excited for this new kind of ‘interactive fan fiction’ and think this new kind of engagement will accrue a lot of value to them, but want the ability to specify how their characters can be used (including not at all).
We assume different people will try very different approaches . . . But we want to apply the same standard towards everyone, and let rightsholders decide how to proceed (our aim of course is to make it so compelling that many people want to).”
Sam Altman, “Sora update #1,” October 3, 2025
Guardrails Up
Testing the app this weekend, I saw just how quickly OpenAI can tighten controls when motivated. Characters from Family Guy, South Park, King of the Hill and numerous other copyrighted properties that I easily generated on Tuesday now trigger content violations. While I was still able to slip a few characters through the filters—Altman admits there will be “edge cases” to work out—the difference is striking.
Which raises an obvious question: If OpenAI could implement these guardrails in 72 hours, why weren’t they there from day one?
The answer may lie in the App Store charts. Despite being invite-only, Sora 2 rocketed to number one within days—fueled by instantly recognizable IP. That doesn’t happen with cautious, permission-first launches; it happens when you allow users to make videos of SpongeBob explaining quantum physics to Peter Griffin. In short, OpenAI used copyrighted characters to drive engagement and media coverage, and it worked.
Perhaps too well.
Buried in Altman’s post is the real story: “People are generating much more than we expected per user,” using massive compute resources on content that’s often being generated for very small audiences (it is, after all, a social media app). OpenAI, Altman wrote, will “have to somehow make money for video generation.” Translation: Sora is burning through compute costs with no offsetting revenue.
When thousands of GPU-intensive ten-second South Park clips also risk copyright lawsuits, the math gets ugly fast. Viewed through this lens, OpenAI’s about-face isn’t contrition; it’s economics. If studios become partners rather than adversaries, OpenAI can potentially offset costs while buying legal cover not only for the outputs, but maybe even retroactive blessing for the unauthorized training as well. It would be a big win for OpenAI, assuming content owners play along.
But that’s a massive if. There’s a world of difference between studios collecting YouTube ad revenue on video clips they produce and control, and handing over their characters for anyone to freely manipulate in exchange for a few bucks.
The next few weeks will reveal whether Hollywood buys into Altman’s “interactive fan fiction” vision. For now, this particular copyright free-for-all appears to be over. The sequel’s probably already in beta.
As always, I’d love to hear your thoughts. Drop a note in the comments below or find me @copyrightlately on social media.
The post Sora, Not Sorry: OpenAI Backtracks on Opt-Out Copyright Policy appeared first on Copyright Lately.