Overview
- CODA submitted a formal written request asking OpenAI to cease using members’ content to train Sora 2 without prior permission and to address infringement claims from member companies.
- Citing reviews of Sora 2 outputs that closely resemble Japanese works, CODA argues reproduction during the machine-learning process may constitute copyright infringement.
- The group rejects OpenAI’s reported opt-out approach for rightsholders, asserting Japan’s system requires advance authorization and offers no shield through later objections.
- CODA represents major publishers and studios including Bandai Namco, Square Enix, Toei, Cygames, and Kadokawa, and was established with government backing to combat piracy and protect cultural exports.
- The Japanese government has already pressed OpenAI to halt alleged infringements, OpenAI has not publicly specified Sora 2’s training data, and legal pressure is mounting globally with U.S. lawsuits proceeding and the Motion Picture Association urging action.