OpenAI has rolled out its new video generation model, Sora 2, with a controversial update that allows it to freely use copyrighted material unless the content and right owners explicitly disallow it.
This ethical detail has stirred a storm within the entertainment industry, pitting the innovations of AI creators against the legal and ethical boundaries of copyright law.
Unlike earlier AI tools requiring explicit permission from copyright owners, Sora 2 was initially released without that option. By default, the new AI-powered video generation model was designed to openly use protected content and generate videos featuring copyrighted characters and other protected content.
This includes copyrighted characters from shows like Spongebob Squarepants, South Park, Pokemon, and Rock and Morty.
OpenAI then included a caveat, the “opt-out” approach, which stated that if rights owners wanted to block Sora 2 from using their data, they must actively submit requests to remove each piece of content or character from OpenAI’s database.
This approach, however, resulted in a fierce backlash from the entertainment and creative industry, including but not limited to Hollywood studios and talent agencies, ultimately raising questions about the protection of intellectual property rights in this AI era.
Within days of the public release, Sora 2-style videos went viral, featuring everything from cartoon characters in illicit scenarios to pop culture mashups. The proliferation of these unlicensed materials using protected content without gaining permission from rights owners set off alarms across the entertainment industry.
Industry Outcry And Legal Concerns
The American Motion Picture Association (MPA), which represents some of Hollywood’s largest studios, led the industry outcry. MPA CEO Charles Rivkin condemned OpenAI for allowing and participating in the unauthorized use of copyrighted works, and also called on the AI company to take “immediate action” to fix the issue.
Rivkin criticized OpenAI for passing the responsibility of copyright enforcement onto holders. “While OpenAI clarified it will ‘soon’ offer rightsholders more control over character generation, they must acknowledge it remains their responsibility – not rightsholders’ – to prevent infringement on the Sora 2 service,” Rivkin said.
Rivkin also warned that the ChatGPT-maker needed to “take immediate and decisive action to address” the proliferation of videos that reportedly infringe the association’s members’ films, shows, and characters.
Major talent agencies like the U.S. Creative Artists Agency (CAA) also joined the opposition and outcry, warning of risks posed by the video-generation model to their clients’ rights. The CAA called Sora 2 a “misuse” of technology that threatens artists’ rights and likenesses.
“CAA is unwavering in our commitment to protect our clients and the integrity of their creations. The misuse of new technologies carries consequences that reach far beyond entertainment and media, posing serious and harmful risks to individuals, businesses, and societies globally. It is clear that Open AI/Sora exposes our clients and their intellectual property to significant risk,” the agency said in a memo.
The CAA further asked if “OpenAI and its partner companies believe that humans, writers, artists, actors, directors, producers, musicians, and athletes deserve to be compensated and credited for the work they create,” or if they believe the company “can just steal it,” by paying little to no attention to global copyright principles and ultimately “dismissing creators’ rights.”
Sam Altman And OpenAI’s Response
Facing mounting pressure and backlash from these agencies, OpenAI CEO Sam Altman acknowledged the copyright issue publicly. In a blog post, Altman emphasized OpenAI’s intention to revert the “opt-out” system and further improve it to allow copyrights have a say in how their data is used.
“We will give rightsholders more granular control over generation of characters, similar to the opt-in model for likeness but with additional controls,” Altman said.
While promising a “very high rate of change” from OpenAI for Sora 2, Altman added that ChatGPT-maker is exploring shifting from the current opt-out to an opt-in model that would require explicit permissions before copyrighted characters appear in AI videos.
Altman also expressed interest in revenue-sharing programs for rights holders who want their intellectual properties featured, aiming to turn the contentious use into a collaborative opportunity between the company and creators.
“People are generating much more than we expected per user, and a lot of videos are being generated for very small audiences,” Altman confirmed. “We are going to try sharing some of this revenue with rightsholders who want their characters generated by users.”
The Bigger Picture: AI and Copyright
This dispute between OpenAI and copyright protectors further highlights the tension between AI innovation and established intellectual property frameworks. Traditional copyright law requires explicit permission or legal exceptions before using protected works.
Against major legal and ethical approaches, OpenAI’s default opt-out approach reverses the usual permission model, making it more likely for the AI-powered video generation model to face scrutiny in courts.
For creators and copyright holders, the stakes are high as unauthorized use of their creations and likenesses threatens revenues and their control over them.
Conversely, OpenAI argues that Sora 2 offers unprecedented creative freedom and fan engagement that aligns with some rights holders that are eager to use and harness AI for growth and new revenue streams. However, the question that remains open and unanswered is how to further balance this AI innovation with copyrights protection.
