OpenAI has been told by Japan’s Content Overseas Distribution Association (CODA) to stop using copyrighted anime and game content to train its AI models. CODA is a major anti-piracy group that includes companies like Studio Ghibli, Bandai Namco, and Square Enix. These studios are known for famous titles like Dragon Ball, Naruto, and Final Fantasy. CODA said OpenAI’s Sora 2 video model creates videos that look too much like Japanese works, which means it might have used their copyrighted content for training without permission.
Japan’s CODA Warns OpenAI Over Sora 2 Copyright Issue
CODA said that many videos made by OpenAI’s Sora 2 look very close to Japanese animation and gaming content. It believes this happened because copyrighted material was used during training. The group added that copying content while training an AI model can still be counted as a copyright violation.
CODA also criticised OpenAI’s earlier “opt-out” policy, where creators had to say they didn’t want their work used. It said Japan’s copyright law doesn’t allow such a system. Instead, companies must get permission before using any copyrighted work. CODA made it clear that OpenAI’s change to an “opt-in” policy came too late.
CODA Demands OpenAI To Stop Using Japanese Content
In its letter, CODA made two strong requests. First, OpenAI must stop using any content made by its member companies for training. Second, the company must give honest answers to questions and complaints from creators and publishers about copyright issues.
CODA said the videos from OpenAI’s Sora 2 are too similar to Japanese content to ignore. It also said that Japan’s rules make it clear, no one can use copyrighted material without permission first.
In September, Japan’s Minister of State for Intellectual Property and AI Strategy, Minoru Kiuchi, also said that the government had asked OpenAI not to break copyright rules. The warning shows how serious Japan is about protecting its creative industry.
