CODA Demands OpenAI Stop Incorporating Ghibli-Style Works into AI Models

Post by : Bianca Suleiman

AI-generated clips that echo Studio Ghibli’s signature visuals have sparked a strong reaction in Japan. The Council of Copyright for Design and Arts (CODA), which speaks for major publishers including Studio Ghibli, has formally asked OpenAI to stop employing copyrighted Japanese material in its model training.

OpenAI’s video tool Sora 2, released on 30 September, rapidly drew attention for creating imagery reminiscent of Ghibli films. CODA wrote to OpenAI on 27 October, asserting that many results "closely resemble Japanese content or images," and suggesting those outputs indicate the company trained its systems on protected works.

Under Japanese copyright statutes, reproducing or leveraging creative works without prior authorization can amount to infringement. CODA criticised OpenAI’s current opt-out approach — which allows creators to remove content only after it has been used in training — saying this practice contravenes Japan’s legal expectation of advance consent.

In its letter, CODA set out two principal demands: that OpenAI immediately halt the use of member works without explicit permission, and that the company engage substantively with copyright concerns raised by the organisation. CODA emphasised the need to reconcile technological progress with the legal and moral rights of creators.

The request follows other controversies around Sora 2. OpenAI recently suspended AI-generated images of Martin Luther King Jr after objections from his estate, underscoring growing scrutiny over content governance and ethical limits for generative systems.

These developments feed into an expanding international debate about AI and intellectual property, as rights holders and regulators press for clearer disclosure, explicit consent mechanisms and greater accountability in how training data is sourced and used.