Japanese Anime Studios Demand OpenAI Halt Sora 2 Training on Copyrighted Work

A major Japanese content group is formally challenging OpenAI. The Content Overseas Distribution Association (CODA) has demanded the AI company stop using its members’ copyrighted anime and film content. The request was made in a letter sent to OpenAI on October 28.

CODA represents prominent Japanese studios, including the legendary Studio Ghibli. The association asserts that OpenAI’s practice of training its Sora 2 video AI on protected material without prior permission violates Japanese copyright law. This move signals a significant escalation in global disputes over AI training data.

Opt-Out Policy Clashes with Japanese Legal Framework

OpenAI’s new Sora 2 model can generate video content featuring popular characters from various media. Instead of preemptively blocking the generation of copyrighted material, OpenAI has implemented an opt-out policy. This policy requires rights holders to object after the AI has already been trained on their work.

According to CODA, this approach is not valid under Japanese law. The group’s letter states that Japan’s copyright system requires prior permission for using copyrighted works. It explicitly notes there is no mechanism to avoid infringement liability through subsequent objections.

Broader Implications for AI Development and Intellectual Property

The conflict highlights a fundamental tension between rapid AI development and established intellectual property rights. CODA’s position is that the very act of replicating copyrighted works during machine learning may itself constitute infringement. This argument goes beyond the output of the AI to challenge the training process itself.

This is not an isolated complaint. Major Hollywood studios have recently voiced similar copyright concerns regarding Sora 2. The growing international pressure from content creators poses a significant challenge to OpenAI’s data sourcing strategies and could reshape how AI models are trained globally.

The standoff between Japanese anime studios and OpenAI over Sora 2 copyright training sets a critical precedent. Its outcome will significantly influence the future of AI development and intellectual property protection worldwide.

Thought you’d like to know

What is CODA?

The Content Overseas Distribution Association is a Japanese organization representing major anime and film studios. Its members include industry giants like Studio Ghibli. CODA works to protect the copyrights of Japanese content abroad.

What did CODA ask OpenAI to do?

CODA sent a formal letter demanding OpenAI stop training its Sora 2 model on members’ copyrighted works. The association insists that prior permission is required under Japanese law, making OpenAI’s opt-out policy non-compliant.

Why is OpenAI’s opt-out policy problematic?

Japanese copyright law generally requires permission before using protected works. CODA argues that asking rights holders to opt-out after training has occurred does not align with this legal requirement and may not shield OpenAI from infringement claims.

How does this affect other AI companies?

This dispute creates a significant precedent. If CODA succeeds, other AI developers training models on copyrighted content may face similar legal challenges. It could force a industry-wide shift towards pre-emptive licensing agreements.

Has Hollywood responded to Sora 2?

Yes, major Hollywood studios have also raised copyright concerns. They have echoed similar points about the need for permission before using protected content for AI training, indicating a growing global consensus among content creators.