Japanese publishers urge OpenAI to stop using their work for training AI

After ChatGPT's image generator was released, a viral trend emerged in which users recreated photos of themselves or their pets in the Ghibli art style and shared them widely online

Don't want to miss the best from Business Standard?

A Japanese trade group called the Content Overseas Distribution Association (CODA), which represents publishers including Studio Ghibli, has written a letter to OpenAI asking it to stop using their copyrighted content to train AI models without permission, according to a report by TechCrunch.

 

The letter comes as OpenAI begins rolling out its new Sora app for video generation. CODA has asked the company not to use its members’ content for machine learning without prior approval and warned that legal action could follow if OpenAI fails to comply.

How did the Studio Ghibli trend spark the dispute?

Studio Ghibli has been especially affected by OpenAI’s image and video tools. After ChatGPT’s image generator was released in March, a viral trend emerged in which users recreated photos of themselves or their pets in the Ghibli art style and shared them widely online. 

 

Even OpenAI CEO Sam Altman joined the trend, changing his profile picture on X to a Ghibli-style version of himself. OpenAI also promoted the trend, which helped boost public interest in its tools.

 

However, Studio Ghibli’s co-founder, Hayao Miyazaki, has long been critical of AI-generated art. According to the report, when shown an AI-created animation in 2016, he said he was "utterly disgusted" and called it "an insult to life itself."

What are CODA’s concerns about copyright and AI?

CODA’s complaint adds to the increasing debate on how OpenAI uses copyrighted material. The company has often followed an "ask forgiveness, not permission" approach, training its AI on copyrighted content first and responding to complaints later.

 

Several major media organisations have already filed lawsuits against OpenAI over similar issues. Here are some of the organisations that have previously sued the company:

• Canadian news publishers: In November 2024, outlets including The Canadian Press, Torstar, Globe and Mail, Postmedia, and CBC/Radio-Canada sued OpenAI, according to the Associated Press.
• The New York Times (US): The paper accused OpenAI of using millions of its articles to train ChatGPT without authorisation. A US federal judge later allowed The New York Times and others to proceed with their lawsuit against OpenAI and Microsoft.
• Alden Global Capital newspapers (US): Eight US newspapers, including the New York Daily News, Chicago Tribune, Orlando Sentinel, Denver Post, and others, also filed a lawsuit alleging unauthorised use of their copyrighted articles to train ChatGPT and Microsoft’s Copilot.

How do copyright laws differ between Japan and the US?

The report added that US copyright law has not been updated since 1976, and there is still no clear legal guidance on whether training AI on copyrighted material counts as infringement. In a recent case, a US judge ruled that AI firm Anthropic did not break copyright law by training its model on copyrighted books, though the company was fined for illegally copying those books for training.

 

According to CODA, Japan’s laws are stricter. The group said that in Japan, using copyrighted works for AI training without prior permission could be considered copyright infringement. CODA said, "In Japan, prior permission is generally required for using copyrighted works. No rule allows someone to avoid breaking the law by asking for approval afterwards."