Japan’s CODA Demands OpenAI Halt Sora 2 Training Over Copyright Concerns

[lingriiddd]

The Content Overseas Distribution Association (CODA) — representing major industry players like Bandai Namco, Square Enix, Studio Ghibli, Universal Music, and Sony Music Entertainment — has issued a formal written request to OpenAI over its Sora 2 text-to-video model.

The request, dated October 27, 2025, urges OpenAI to cease training Sora 2, launched on September 30, citing evidence that the model’s outputs “closely resemble Japanese content or images.” CODA alleges this stems from the unauthorized use of Japanese media in training data, raising potential copyright infringement concerns.

[EmTeam]

Fair use vs. copyright — this battle is far from over. ️

[jacson4]

Transparency in training data is becoming the new regulatory frontier.