OpenAI is facing scrutiny over its Sora 2 application, which has generated a flurry of videos featuring copyrighted characters, including beloved Nintendo icons. In response, OpenAI has pledged to offer copyright holders “more granular control” over how their intellectual property is used within the app.
Following the debut of Sora 2, social media platforms were inundated with AI-generated videos showcasing familiar licensed characters like Mario, Pikachu, and a host of other Pokémon. This prompted widespread discussion and concerns among IP owners.
Before Sora 2’s official launch, a report by the Wall Street Journal indicated that OpenAI had already engaged with major movie studios and other intellectual property owners. The company offered them an opportunity to opt out their fictional characters from being used in AI-generated videos. This approach differed significantly from how OpenAI handled the likeness of public figures, which required an opt-in consent.
However, OpenAI CEO Sam Altman later shared in a blog post that the company was actively “taking feedback” from users, rights holders, and various other groups, signaling upcoming changes. He confirmed that IP owners would gain “more granular control” over character generation, similar to how individuals can opt-in to share their own likeness, but with additional safeguards.
Altman noted that many rights holders expressed excitement for this new form of “interactive fan fiction,” believing it could generate significant value for them. However, they also emphasized the need to specify how their characters could be used, including the option to disallow their use entirely.
OpenAI CEO Sam Altman has referred to Sora 2 videos featuring copyrighted characters as “interactive fan fiction.”
Altman further stated that OpenAI would “let rights holders decide how to proceed” and acknowledged that some “edge cases” of generations might bypass the platform’s protective guardrails. He also hinted at future monetization strategies for video generation, suggesting that payments could be made to rights holders who grant permission for their characters to be used. “The exact model will take some trial and error to figure out, but we plan to start very soon,” Altman explained. “Our hope is that the new kind of engagement is even more valuable than the revenue share, but of course we want both to be valuable.”
He predicted a rapid pace of change, reminiscent of the early days of ChatGPT, and anticipated both “good decisions and some missteps” along the way.
Significantly, Altman highlighted Japan’s unique creative landscape, expressing admiration: “In particular, we’d like to acknowledge the remarkable creative output of Japan — we are struck by how deep the connection between users and Japanese content is!”
Altman’s comments coincide with an unusual official statement from Nintendo, which appeared to be a direct rebuttal to remarks made by Satoshi Asano, a member of Japan’s House of Representatives. Asano had previously accused Nintendo in a deleted social media post of “avoiding using generative AI to protect its IP” and “engaging in lobbying activities with the government” concerning the broader use of generative AI in the gaming industry.
Nintendo firmly refuted these claims, asserting that it had “not had any contact with the Japanese government about generative AI.” The company reiterated its commitment to taking “necessary actions against infringement of our intellectual property rights,” regardless of whether generative AI is involved.
Adding to the conversation, CNBC reported seeing a Sora 2 video where Sam Altman appeared alongside several Pokémon characters, humorously stating, “I hope Nintendo doesn’t sue us.”
This incident reflects a broader trend; Disney and Universal have already initiated lawsuits against AI image creator Midjourney, alleging unauthorized use and distribution of AI-generated characters from their films. Disney also issued a cease and desist order to Character.AI for similar reasons.
Mark Lemley, a Stanford Law School professor, warned that “A lot of the videos that people are going to generate of these cartoon characters are going to infringe copyright,” suggesting that “OpenAI is opening itself up to quite a lot of copyright lawsuits by doing this.”
In a discussion with IGN, business lawyer Richard Hoeg, host of the Virtual Legality podcast, admitted that the legal standing of OpenAI’s actions remains ambiguous. “In short, we don’t have a definitive answer yet,” he stated. Hoeg elaborated that while training AI on protected materials might be considered legal if the materials were lawfully obtained, the output side presents more challenges. He cited the Disney/Dreamworks lawsuit, which argues that platforms should actively police prompts for infringing content, especially given their existing content moderation for other issues.
Hoeg concluded, “The law moves slowly, far slower than technology, which is why you see these tech companies racing ahead of it a bit. My best guess is that OpenAI is probably going to be okay long term on the training sets they used (assuming they weren’t pirated), and that the ‘opt out of training’ option therefore won’t do much of anything. Where they really need to concern themselves is on the output side and/or if they are marketing their software’s abilities with protected content themselves.”
IGN has reached out to Nintendo and The Pokémon Company for official comments on these developments.
This follows a prior incident where The Pokémon Company responded to the Department of Homeland Security’s unauthorized use of Ash Ketchum and the Pokémon theme song in a video. A spokesperson clarified that “Our company was not involved in the creation or distribution of this content, and permission was not granted for the use of our intellectual property.”
This article was written by Wesley.