🔗 Share this article

The company is promising to offer rights owners with increased oversight over the creation of characters after its newly launched tool, Sora 2, generated a wave of clips depicting copyrighted characters.

Sora 2’s Release and Initial Issues

The AI video generator, driven by AI, was rolled out recently on an exclusive basis. The application enables individuals to produce brief clips from a written description. Initial assessments of the machine-created video feed revealed copyrighted characters from well-loved series including animated series, South Park, Pokémon and Rick and Morty.

Prior to the release of Sora 2, OpenAI reportedly told talent agencies and production houses that if they wished to avoid their copyrighted material replicated by the video generator, they would need to decline.

“We’ll work with rights holders to block characters from Sora at their request and respond to takedown requests.”

The company explained that rights holders can flag copyright infringement via a “copyright disputes form”, but individual artists or companies are not granted a blanket opt-out.

CEO’s Response and Upcoming Plans

Recently, OpenAI CEO Sam Altman said in a blog post that the company had been “taking feedback” from users, rights holders and other groups, and would make adjustments as a result.

He mentioned that rights holders would be provided more “granular control” over character generation, akin to how users can choose to use their own image in the application, but with extra safeguards.

“Many copyright owners have expressed enthusiasm for this innovative form of ‘interactive fan fiction’ and believe it will generate significant value, yet they desire the power to define usage terms for their characters, including complete prohibition.”

Addressing Edge Cases and Revenue Plans

Altman said that OpenAI would allow copyright owners to determine their approach and that there would be some unusual instances of content that bypass the system’s protections that should not.

He added that the company would also need to generate revenue from content creation, and currently the system was observing people producing far greater content than anticipated per user. He suggested compensation to rights holders who allow their figures to be generated.

“The exact model will take some trial and error to figure out, but we plan to start very soon,” he commented. “We anticipate that this novel interaction will surpass monetary gains in importance, though we aim for both aspects to be beneficial.”

He concluded by noting that there would be a rapid pace of evolution, comparable to the initial stages of ChatGPT, and there would be both successful choices and occasional errors.