The Content Overseas Distribution Association (CODA), representing Japanese IP holders such as Studio Ghibli and Bandai Namco, has formally requested that OpenAI stop using its members’ content to train Sora 2. The organization argues that replication during the machine learning process may constitute copyright infringement, as Sora 2 generated content featuring copyrighted characters after its September 30 launch.
CODA’s concerns follow similar incidents, including the proliferation of “Ghibli-style” images after the GPT-4o launch in March. Japan’s government has also urged OpenAI to halt replication of Japanese artwork. While OpenAI announced last month that it would implement an opt-out policy for IP holders, CODA contends that Japanese copyright law generally requires prior permission, meaning subsequent objections may not shield the company from liability.
CODA is requesting that OpenAI “responds sincerely” to copyright claims and cease using its members’ content in machine learning processes. The notice addresses not only Sora 2 outputs but also the underlying use of Japanese IP as training data. OpenAI has yet to publicly comment on the organization’s formal request.
