Japan’s CODA urges OpenAI to end unauthorized training on copyrighted content, spotlighting growing global tensions over AI and intellectual property.
A Clear Message from Japan’s Creative Industry
The Content Overseas Distribution Association (CODA) — a Japanese trade body representing major content producers like Studio Ghibli — has formally requested OpenAI to stop using their copyrighted works for AI model training without permission.
- CODA’s letter specifically calls out the unauthorized use of creative assets in machine learning workflows.
- The letter follows widespread online use of Ghibli-style AI art, especially after the release of OpenAI’s image and video tools.
Among the most visible signs of this trend was Sam Altman, OpenAI’s CEO, changing his X profile picture to a stylized, AI-generated “Ghiblified” portrait — a move symbolic of the broader cultural appropriation CODA is pushing back against.
AI Style Mimicry Raises Legal and Ethical Questions
OpenAI’s generative models have enabled users to easily recreate selfies, pets, or characters in the distinct visual style of Ghibli films like Spirited Away or My Neighbor Totoro.
- These tools, including the upcoming Sora video generator, replicate copyrighted aesthetics, raising concerns of cultural dilution and artistic exploitation.
- CODA warns that even training AI on protected works without permission may constitute copyright infringement under Japanese law.
In Japan, prior authorization is mandatory for the use of copyrighted materials, and post-facto objections do not absolve liability — a stance far stricter than prevailing U.S. norms.
Legal Grey Zones: U.S. vs. Japan
OpenAI’s legal strategy thus far has leaned on a “use first, address complaints later” approach, citing fair use doctrines under U.S. copyright law.
- However, copyright law in the U.S. has not been significantly updated since 1976, leaving many questions unresolved.
- A recent case involving Anthropic illustrated this legal ambiguity: a federal judge ruled that training on copyrighted books wasn’t illegal — though the company was penalized for acquiring the materials illegally.
In contrast, Japan’s CODA insists that replication during training may already breach copyright, especially if the model later generates recognizably similar output.
Ghibli’s Founder Already Voiced Concerns
While Hayao Miyazaki, the creative force behind Ghibli, hasn’t commented on this latest controversy, his stance on AI art is well documented.
- In 2016, when shown an AI-generated 3D animation, Miyazaki said he was “utterly disgusted,” calling the technology an “insult to life itself.”
- His views reflect a deeper discomfort in the Japanese animation and art community, where emotional and hand-crafted storytelling is held in high regard.
This philosophical clash — between machine-generated content and human creativity — sits at the heart of CODA’s objection.
A Global Copyright Reckoning Is Looming
As AI companies like OpenAI scale their tools, pressure from international rights holders is intensifying.
- Nintendo, the estate of Dr. Martin Luther King Jr., and now Studio Ghibli’s industry allies have expressed concern over deepfakes and AI mimicry.
- CODA’s position reflects a broader international movement seeking explicit consent and compensation for the use of creative IP in AI systems.
Whether OpenAI complies voluntarily or faces legal action remains to be seen. But the global patchwork of copyright laws could eventually force AI companies to adapt region-specific training protocols or face regulatory and reputational blowback.
Japanese publishers, including Studio Ghibli, are calling on OpenAI to stop using their copyrighted content to train AI models without permission. With Japan’s stricter copyright laws and growing global concern over AI-generated media, the pushback signals a rising demand for consent-based model training.
