Japanese Government Calls on OpenAI to Refrain from Copyright Infringement (www.ign.com)

Japan’s Cabinet Office has formally asked OpenAI to refrain from copyright infringement after Sora 2 — OpenAI’s new video generator launched Oct. 1 — produced 20‑second, 1080p clips with sound that widely circulated online and often depicted copyrighted anime and game characters (One Piece, Demon Slayer, Pokémon, Mario, etc.). Officials including Minister Minoru Kiuchi and Digital Minister Masaaki Taira urged voluntary compliance and warned that measures under Japan’s AI Promotion Act (fully enforced Sept. 1, 2025) could be invoked; the act empowers analysis and countermeasures for AI harms but does not set explicit penalties. OpenAI CEO Sam Altman said the company will add more granular, opt‑in/opt‑out controls so rightsholders can specify or block character uses, acknowledging “edge cases” will need fixing. Nintendo and other rights holders have signaled readiness to take action, while U.S. studios have already sued or threatened AI firms over character use. For the AI/ML community this crystallizes practical and legal constraints on generative models that produce copyrighted likenesses: product teams must implement fine‑grained rights controls, provenance and usage policies, detection/watermarking, and licensing flows to mitigate infringement risk. The episode highlights cross‑border regulatory pressure, potential litigation exposure (as seen in recent suits vs. image‑model vendors), and an emergent expectation that platforms offer technical levers for rights management—shaping how multimodal models are designed, deployed, and commercialized.