You can’t use copyrighted characters in OpenAI’s Sora anymore and people are freaking out

The full copyright-for-all approach taken by OpenAI in its new AI-driven video creation model, Sora 2, ran for one week. After initially asking copyright holders to opt out of having their content appear in videos created by Sora, CEO Sam Altman Announce That the company will move to a “subscription” model that “gives rights holders more granular control over character creation” — and Sora’s obsession doesn’t take well.

Given the type of content created with Sora and shared via the TikTok-style social app that OpenAI launched specifically to host user-generated Sora videos, the change shouldn’t come as a shock. Almost immediately, it was the platform Flooded with copyrighted material They’re being used in ways that the rights holders almost certainly didn’t care about, unless you think Nickelodeon really liked subversion Nazi Spongebob. On Monday, the Motion Picture Association became one of the loudest Contact to OpenAI to put an end to the potential infringement. It didn’t take long for OpenAI to respond and comply.

In a blog post, Altman said Sora’s new approach to copyrighted material will require rights holders to opt in to the use of their characters and content — but he’s pretty sure copyright holders love the videos, in fact. “We’re hearing from a lot of rights holders who are very excited about this new kind of ‘interactive fan fiction’ and think this new kind of engagement will bring them a lot of value, but they want the ability to decide how their characters will be used (including not using them at all),” Altman wrote, noting that his company wants to “let rights holders decide how to move forward.”

Altman also admitted, “There may be some generational edge cases that shouldn’t be skipped, and getting our collection to run well will require some iteration.” It’s not clear whether that will play out with rights holders. MPA CEO Charles Rifkin said in a… statement OpenAI “must acknowledge that it remains its responsibility – not rights holders’ – to prevent infringement on the Sora 2 service,” it said, “and well-established copyright law protects the rights of creators and applies here.”

While OpenAI may give copyright holders more control over the output of its model, they don’t seem to have much say over the inputs. Report from The Washington Post showed How the first version of Sora was explicitly trained on copyrighted material that the company had not asked permission to use. It’s not clear if OpenAI went out and obtained those rights to train Sora 2, but the generator is very good at producing exact copies of copyrighted material in a way that it can only do if it is fed a large trove of existing content during training.

The largest AI training case to date Anthropists pays $1.5 billion To settle a copyright infringement case with the authors of books the company pirated to train its models. The judge in this case found so Using copyrighted material for training without permission is fair usealthough other courts may disagree with this call. Earlier this year, OpenAI asked the Trump administration To call fair use for training an AI model. So a lot of OpenAI’s strategy around Sora seems to be manipulating and hoping that, if she makes the right allies, she’ll never have to find out.

OpenAI may have been able to appease copyright holders by changing its Sora policies, but it is now angering its users. like 404 reported by the mediasocial channels like Twitter and Reddit are now full of Sora users angry that they can no longer create 10-second clips featuring their favorite characters anymore. One user in OpenAI subreddit He said that being able to play with copyrighted material was “the only reason this app was so fun.” Another claimed that “moral policing and left-wing ideology are destroying America’s AI industry.” So, you know, they seem to be handling this well.