The Content Overseas Distribution Association (CODA), representing major entities such as Studio Ghibli and Square Enix in the anime and video game sectors, has issued a formal request to OpenAI to refrain from unauthorized utilization of its intellectual property in the Sora 2 AI video generator. This request was communicated through a letter sent on October 27 and was made public the following day, as reported by Automaton. CODA asserts that the replication during the process of machine learning may potentially violate copyright laws.
Representing 36 significant Japanese firms, including industry giants like Aniplex, Toei, and Universal Music, CODA points out that a significant amount of Sora 2’s generated content closely resembles Japanese works or images. This similarity is attributed to the input of Japanese materials during the machine learning phase.
Concerns from Japanese Officials
In related developments, officials from the Japanese government have labeled manga and anime as “irreplaceable treasures.” They have urged OpenAI to act against copyright violations, emphasizing that Japan must “take the lead” in regulating AI technologies.
Backdoor Angles and Nintendo’s Response
Some former marketing executives from Nintendo speculate that the company may seek alternative approaches while pursuing legal action against AI developers for generating counterfeit Pokémon content, drawing parallels with the game Palworld.
Nintendo has clarified that it has had no discussions with Japanese authorities regarding generative AI, yet it remains committed to combating any infringement of its intellectual property.
Japan’s Copyright Landscape
The structure of Japan’s copyright laws necessitates obtaining permission in advance to use copyrighted content. This contrasts with Sora 2’s current model, which allows companies to opt out of featuring certain character likenesses only midway through their use. CODA has cautioned OpenAI regarding this issue, stressing that there is no framework that permits individuals to evade liability for infringement through later objections.
Response from OpenAI’s CEO
In a blog post dated October 3, OpenAI CEO Sam Altman addressed the concerns surrounding the unpopular opt-out policy of Sora 2. He mentioned that the service will eventually provide content creators with more detailed control over character generation, akin to an opt-in framework for likeness but supplemented with further options.
Altman expressed admiration for Japan’s significant creative output, stating that the profound bond between users and Japanese media is impressive. He indicated intentions to share some of the revenue with rightsholders who approve of their characters being generated by users.
CODA’s Formal Requests
However, CODA finds these measures insufficient. The organization has made two specific requests to OpenAI: first, that their members’ content is not utilized for machine learning without prior consent; second, that OpenAI addresses claims and inquiries from CODA’s member companies regarding potential copyright infringements related to Sora 2’s results.
CODA emphasizes the need for genuine engagement, stating that it is crucial for fostering both the progression of AI technology and the safeguarding of the rights of creators and rightsholders.
In another note, Take-Two’s CEO recently commented that AI will likely struggle to effectively create video games and might not even devise a plausible marketing strategy for the upcoming GTA 6.