Studio Ghibli, Bandai Namco, Square Enix, and other Japanese rights holders have asked OpenAI to stop using their content to train Sora 2. The request came via CODA, Japan’s Content Overseas Distribution Association, which issued a public letter on October 27 arguing that replication during machine learning may constitute copyright infringement under Japanese law. CODA also objects to Sora 2’s opt-out approach for IP owners. Under Japan’s system, prior permission is generally required and later objections do not erase liability.
This follows the late-September release of Sora 2, which triggered a flood of videos and images that resemble well-known Japanese properties. Japan’s government has since asked OpenAI to seek permission from rights holders before allowing such replications, framing manga and anime characters as cultural assets that require protection.
The IP regime in Japan privileges explicit, opt-in licensing. An opt-out control may not mitigate exposure in that market. If your brand assets, franchises, or characters are valuable in Japan, your legal position depends on proactive terms, not platform toggles. Cross-border compliance will require different strategies for the U.S., the EU, and Japan. Model providers are iterating policies, but enforcement and liability will hinge on jurisdiction, contract language, and actual outputs.
This is obviously a question for your GC, but as best I can tell, we need to treat Japan as an opt-in territory for AI training and generation, which means that for any AI content program that touches Japan, you’re going to require vendor warranties and indemnities that reflect prior-permission standards. This is going to be a pain, because we’re going to have to map where assets are likely to be replicated and set technical blocks wherever it’s possible to do so. Oh, and add another territory to your takedowns and claims workflows. I wonder how many other countries are going to create a mountain of meta-work with their unique AI policies? That would be all of them. Ugh!
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.