Studio Ghibli and Japanese Publishers Challenge OpenAI Over Unauthorized AI Training

Ghiblification, generative AI and the quest for fair circuits of creativity - Nature Electronics

Spurred by a new image-generation feature embedded in OpenAI's generative pre-trained transformer 4 omni (GPT-4o) model, artificial intelligence (AI)-generated images in the style of Studio Ghibli spread across social media in late March 20251. These images -- characterized by soft watercolour skies and the lyrical stillness perfected by Japan's Studio Ghibli -- were generated primarily using text prompts by third-party users. The incident demonstrated how a single text prompt could replicate a globally recognized artistic style, while navigating the periphery of conventional copyright law. The controversy sparked an outcry among artists and fans, who argued that even if an artist's style has no formal copyright protection, using it at scale via AI without permission is unethical. This Ghiblification phenomenon crystallized a global conflict over the legal and ethical status of artistic style, a conflict that unfolds across different legal systems with differing frameworks. Japan, for instance, adopts a pro-innovation strategy of deliberate ambiguity. Japan's 2018 Copyright Act amendment (Article 30-4) grants broad exemptions for AI training for 'non-enjoyment' purposes, aiming to attract AI development. However, this is restricted by the 2024 government guidance, which prohibits uses that replicate specific works or unreasonably prejudice copyright holders. This drives companies towards pre-emptive licensing and self-regulation, reflecting a top-down model aligned with industrial policy and social harmony. In the USA, the longstanding idea-expression dichotomy implies that an artist's style or aesthetic cannot be directly copyrighted as it is considered an unprotectable idea or method rather than protectable expression (17 U.S.C. § 102(b)). Consequently, artists concerned about AI models replicating their style have no specific statutory remedy available under current law. The pending class-action Andersen v. Stability AI exemplifies this conflict: the plaintiffs allege that the Stable Diffusion model was trained on billions of scraped images (including their own) without permission and that its model parameters (weights) embed unauthorized reproductions of their art, effectively making the AI system an infringing derivative work. In response, the defendants argue that no protectable expression was copied, claiming the AI merely learns general artistic 'style' (which is not copyrightable by law) and generates new images in a transformative manner analogous to a human artist drawing inspiration from others. The crux of the dispute is whether the court would accept this fair-use analogy or side with the plaintiffs' view that generative AI serves as a market substitute for their creative output. Notably, the US Copyright Office's 2025 report has echoed the plaintiffs' concerns, warning that the proliferation of AI-generated works in original artists' styles could flood the market and dilute the value of their work. In South Korea, digital comic (webtoon) artists became concerned that Naver -- a leading AI company and developer of the HyperCLOVA large language model suite -- or other webtoon artists were using their work without permission to train AI models. Naver hosts one of the largest online comic platforms (Naver Webtoon) in the country. In mid-2023, frustrated by the lack of legal protection, more than 200 webtoon artists publicly declared their intent to join a creative online protest on Naver Webtoon. The protest took place in Naver's Challenge Comics section (an open forum for amateur creators), which was flooded with more than 60 newly uploaded comics titled AI Webtoon Boycott, each prominently displaying an AI-ban icon. The campaign was effective, quickly attracting public attention and prompting a corporate response. Naver banned AI-generated works from its contests and assured artists their content would not be used for AI training. Global AI firms such as OpenAI must navigate this fragmented legal landscape, resulting in inconsistent, risk-driven policies. OpenAI's decision to block prompts referencing living artists, while allowing corporate styles such as Studio Ghibli's, reflects this approach. Lawsuits by individuals may invoke the right of publicity, a clearer and more enforceable claim in the USA than copyright protections for 'style'. This legal risk compels private companies to act as de facto global regulators of creative expression. Their policies, shaped by legal exposure and public pressure, set precedents traditionally defined by legislation. The Ghibli controversy highlights the absence of a unified solution to generative AI's impact on artistic identity. Japan's administrative approach promotes innovation but depends on corporate self-restraint. The US litigation-based model may yield legal clarity but is slow and costly. South Korea's activist model empowers creators but remains vulnerable to corporate circumvention. A sustainable framework must integrate these strengths: self-regulation and licensing incentives, legal clarity from litigation, and creator-led collective action.

Studio Ghibli and other Japanese publishers want OpenAI to stop training on their work | TechCrunch

A Japanese trade organization representing publishers like Studio Ghibli wrote a letter to OpenAI last week, calling for the AI giant to stop training its AI models on their copyrighted content without permission. Studio Ghibli, the animation studio behind films like "Spirited Away" and "My Neighbor Totoro," has been especially impacted by OpenAI's generative AI products. When ChatGPT's native image generator was released in March, it became a popular trend for users to prompt for recreations of their selfies or pet pictures in the style of the studio's films. Even OpenAI CEO Sam Altman changed his profile picture on X to a "Ghiblified" picture. Now, as more people get access to OpenAI's Sora app and video generator, Japan's Content Overseas Distribution Association (CODA) has requested that OpenAI refrain from using its members' content for machine learning without permission. This request does not come unprompted. OpenAI's approach to working with copyrighted content is to ask forgiveness, not permission, which has made it all too easy for users to generate photos and videos of copyrighted characters and deceased celebrities. This approach has yielded complaints from institutions like Nintendo, as well as the estate of Dr. Martin Luther King, Jr., who could very easily be deepfaked on the Sora app. It's up to OpenAI to choose whether or not to cooperate with these requests; if not, the aggrieved parties can file a lawsuit, though United States law remains unclear about the use of copyrighted material for AI training. There is little precedent thus far to guide judges on their interpretation of copyright law, which has not been updated since 1976. However, a recent ruling by US federal judge William Alsup found that Anthropic did not violate the law by training its AI on copyrighted books -- the company did get fined for pirating the books it used for training, though. But Japan's Content Overseas Distribution Association (CODA) claims that this may be considered a copyright violation in Japan. "In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement," CODA wrote. "Under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections." Hayao Miyazaki, one of the central creative figures of Studio Ghibli, has not commented directly on the proliferation of AI-generated interpretations of his work. However, when he was shown AI-generated 3D animation in 2016, he responded that he was "utterly disgusted." "I can't watch this stuff and find it interesting," he said at the time. "I feel strongly that this is an insult to life itself."

A Trade Group That Includes Studio Ghibli Just Slapped OpenAI withâ€| a Letter

A Japanese trade organization that includes heavy-hitting media creators like Studio Ghibli, Square Enix, and Bandai just announced that it sent a letter to OpenAI dated October 28 concerning alleged copyright violations. The letter includes some observations about the similarity of Sora 2 videos to “Japanese content,†and issues two requests: It asks OpenAI not to use CODA content as training data without prior permission, and requests that OpenAI “responds sincerely†when a CODA member complains about copyright issues. Notably absent are anything like “demands†of “immediate action,†or any sort of direct legal threats. Sora 2, OpenAI’s top-of-the line text-to-video model was released in late September, and anyone with an interest in AI watched in a mix of amazement and disgust as copyright hell was unleashed immediately. That included a great deal of content that looked a lot like Japanese media properties like Pokemon, Hideo Kojima’s video game universes, and some unspecified Studio Ghibli production. The framing of the alleged infringement is different in tone and approach than most American copyright claims. The similarity between Sora 2 and Japanese images and video “is the result of using Japanese content as machine learning data,†CODA says. When such content is the output, “CODA considers that the act of replication during the machine learning process may constitute copyright infringement.†Japan’s Copyright Act has a potentially relevant section on AI called Article 30-4 that may shed some light on CODA’s logic, and its reason for starting with such a gentle approach to achieving redressâ€"namely that Japan is a permissive legal environment for this sort of thing. According to a government fact sheet on the law, “exploitation for non-enjoyment purposes†such as “AI development or other forms of data analysis may, in principle, be allowed without the permission of the copyright holder.†CODA, however, says that in Japan, “prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections.â€

Studio Ghibli, Square Enix demand OpenAI stop training Sora 2 on their content

A group representing Japanese animation companies -- notably, the widely beloved Studio Ghibli -- has demanded OpenAI stop using their art to train Sora 2, its AI video tool. The trade group called the Content Overseas Distribution Association (CODA) released a statement this week, stating that it had submitted a request to OpenAI at the behest of its member companies. Wrote CODA: "CODA has confirmed that a large portion of content produced by Sora 2 closely resembles Japanese content or images. CODA has determined that this is the result of using Japanese content as machine learning data. In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement. Furthermore, according to media reports, it has been stated that Sora 2 responds through an opt-out system based on requests from copyright holders. However, under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections." CODA requested that OpenAI stop using the art of Japanese animation companies without permission and that the company "responds sincerely to claims and inquiries from CODA member companies regarding copyright infringement." CODA represents Studio Ghibli, video game publisher Square Enix (maker of Final Fantasy), and many other publishers. No company has had a more checkered history with OpenAI than Studio Ghibli, however. A viral trend in March had folks creating Ghibli-style images and videos via ChatGPT and Sora. That proved to be widely controversial, considering that beloved Studio Ghibli co-founder Hayao Miyazaki is vehemently opposed to AI. "Whoever creates this stuff has no idea what pain is whatsoever," he said in a 2016 documentary. "I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it. I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself." That didn't stop OpenAI from fully embracing the trend, however. Now, months later, it appears the fight continues.

Studio Ghibli Demands That OpenAI Stop Ripping Off Its Work

After Sora 2 was used to relentlessly churn out depictions of Japanese anime and video game characters, the creators of those characters are striking back. On October 28, a group representing Studio Ghibli, Bandai Namco, Square Enix, and other major Japanese publishers submitted a written request to OpenAI demanding that it stop using their copyrighted content to train the video generating AI tool. The move, as first reported by Automaton, is the latest example of Japan signaling protectiveness of its art and media against an AI industry that catapulted itself to extraordinary heights by devouring copyrighted works en masse without permission or compensation. In a statement, the group, called the Content Overseas Distribution Association (CODA), said that it had determined Sora 2 is able to generate outputs that "closely resembles Japanese content or images" because this content was used as training data. Therefore, in cases "where specific copyrighted works are reproduced or similarly generated as outputs," CODA said, it "considers that the act of replication during the machine learning process may constitute copyright infringement." The launch of Sora 2 proved to be yet another generative AI tool with an irreverent attitude toward copyright law, only one that was designed to instantly feed into an endless TikTok-style scroll of short-form vertical videos. Here, there was no pretending that OpenAI's goal was to do anything other than serve up slop. Recognizable characters like SpongeBob were often parodied with the AI, and perhaps none more than Japanese characters across various franchises. Many Sora videos featured Pokemon, including one showing a deepfaked OpenAI CEO Sam Altman grilling a dead Pikachu, and another showing Altman gazing at a flock of Pokemon frolicking across a field, before grimacing into the camera and saying, "I hope Nintendo doesn't sue us." The real Sam Altman, in fact, has acknowledged his fans' affinity for ripping off Japanese art, though without mentioning the outrage this caused. "In particular, we'd like to acknowledge the remarkable creative output of Japan," he wrote in a blog post following the launch of Sora 2."We are struck by how deep the connection between users and Japanese content is!" This "deep" connection goes back a while. When OpenAI released new image generation capabilities for ChatGPT in March, it spawned a mega-viral trend of using the tool to generate images imitating the style of the legendary Japanese animation house Studio Ghibli, including "Ghiblified" selfies of yourself. Altman consecrated the trend by creating his own Ghibli-style portrait, which is still his profile picture on social media to this day. It's no surprise, then, that Japan has been feeling a little on edge about OpenAI's attitude towards copyright. Mid-October, the Japanese government made a formal request asking it to stop ripping off the nation's beloved characters. Minoru Kiuchi, the minister of state for IP and AI strategy, called manga and anime "irreplaceable treasures." Initially, OpenAI signaled that copyright holders would have to manually opt out of having their works cribbed by Sora, but then reversed course after the launch and said that they would be opted out by default. Crucially, this was only after it piggybacked the virality of using copyrighted characters to shoot the app to the top of Apple's App store. CODA also doesn't seem to consider this a satisfactory measure. In the statement, the group notes that "under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections." The CODA members have requested the following: that their content isn't used for AI training without permission, and that OpenAI "responds sincerely to claims and inquiries from CODA member companies regarding copyright infringement related to Sora 2's outputs."

Japan's biggest publishing house and one of the world's largest manga producers takes aim at OpenAI, accusing it of 'trampling on the dignity of artists'

Copyright is pretty big to a major publisher, in case you can't tell. You may not know the name Shueisha off the top of your head, but you likely know the works it is responsible for. Under its Jump line of magazines, Shueisha published One Piece, Dragon Ball, Demon Slayer, and recent mega hit Chainsaw Man. This is all to say that Japan's biggest publishing house owes its success to the works of hundreds of works and their creators. It seems OpenAI didn't quite get this memo, as it's being accused of infringing copyright. Shueisha recently put out a statement aimed at OpenAI, and specifically its Sora 2 video generator. At the end of September, we got a glimpse of Sora 2, and one of the first viral posts from this tool was deemed "Temu Blue Exorcist", which is a reference to another manga published by Shueisha. This isn't the first time OpenAI's AI generation tools were used to ape anime and manga, with OpenAI CEO Sam Altman even sharing a picture of himself in Studio Ghibli style. As of the time of writing, Altman's profile picture on X is still in Ghibli style. That statement says (machine translated): "These videos, which infringe on copyrights for anime and characters, were generated using AI learning. Advances in generative AI should be welcomed as they enable more people to share the joy of creation and enjoy creative works. However, this should not be achieved by trampling on the dignity of artists who poured their heart and soul into their world and violating the rights of many others." The complaint argues that generative AI tools should give creators broader and more effective measures to avoid having their work scraped, other than a single opt-out system. Opt-out systems often get ignored and don't adequately give informed consent to parties whose works are being used. Shushei continues, "Regardless of whether generative AI is used or not, we will take appropriate and strict action against anything that we determine to be infringing the rights related to our works" The world's biggest manga publisher isn't without its criticism when it comes to the artists it relies on. Being a big publishing house, it can make many young mangaka (manga creators), but, as reported by The New Yorker, it's gruelling work and "akin to running a marathon at a sprinter's pace". In the New Yorker's interview with Takeru Hokazono, the creator of Kagurabachi, Hokazono revealed he would only take one evening a week to himself, and that's an expectation in the wider industry. Despite Shushei offering mental and physical support to workers, the manga industry at large is known for having a burnout problem. Shushei is just one of many companies to issue such warnings. CODA, the Japanese industry group with members such as Square Enix, Bandai Namco, From Software, and even Studio Ghibli, has issued a statement that "a large portion of content produced by Sora 2 closely resembles Japanese content or images." It argues that the "act of replication during the machine learning process may constitute copyright infringement." The roots of PC gaming have taken liberally from manga. Dark Souls creator Hidetaka Miyazaki is well known for his love of Berserk, Akira is known for its importance to the Cyberpunk genre (I mean, just look at how iconic that bike drift is), and classic fighting manga like Dragon Ball not only have tonnes of games but are influential in all kinds of media. Whatever measures Shueisha may take from here will be done "in cooperation with copyright holders and related organisations to build and maintain a sustainable creative environment."

Japanese publishers like Bandai Namco and Square Enix are requesting Sora 2 is no longer trained on their creative works

A collection of Japanese publishers including Bandai Namco, Square Enix, and others have come together as part of CODA (Content Overseas Distribution Association) to request Open AI cease training AI video generation tool Sora 2 on their collective creative works. The statement to Open AI, made public via a statement published on the CODA website, state that CODA, requests that CODA members works are not used for machine learning without prior permission, and that Open AI responds sincerely to claims and inquires from CODA members regarding copyright infringement related to Sora 2's output. The statement also mentions Sora 2's opt out policy, that being that copyrighted works will be used by Sora 2 unless copyright owners specifically opt out themselves. Coda emphasises that this goes against Japanese copyright infringement laws in which permission is required beforehand. Coda also noted that: "there is no system allowing one to avoid liability for infringement through subsequent objections." Coda is a substantial organisation in Japan that is responsible for copyright protection through combating piracy, as well as the legal worldwide distribution of Japanese video games, movies, music, T.V, and animation. It has a wide variety of members, including companies like Bandai Namco, Cygames, Toei, Square Enix, and more. Coda is among many with serious copyright concerns in regard to AI models such as Sora 2. AI company Anthropic agreed to pay $1.5bn to authors to settle a copyright lawsuit earlier this year, while a wide variety of ongoing lawsuits are currently in the works (all of which can be tracked on this handy Wired article)

Square Enix and other game studios issue notice to OpenAI to stop using their content for AI

TL;DR: The Content Overseas Distribution Association (CODA), representing major Japanese game studios, has formally requested OpenAI address copyright concerns over its Sora 2 AI video generator. CODA alleges unauthorized use of Japanese content for training, emphasizing that replication during machine learning may infringe copyrights under Japan's laws. The Content Overseas Distribution Association (CODA) is a Japanese anti-piracy and copyright-related organization that represents a large number of game development studios and publishers, including Square Enix, Bandai Namco, and FromSoftware - creators of some of the most influential and recognized gaming franchises. Final Fantasy, Dragon Quest, Tekken, Pac-Man, Dark Souls, Elden Ring, and more. And on October 27, CODA submitted a written request to OpenAI regarding its Sora 2 AI video generator. "CODA has confirmed that a large portion of content produced by Sora 2 closely resembles Japanese content or images," the announcement says. "CODA has determined that this is the result of using Japanese content as machine learning data." The organization is requesting that OpenAI respond "sincerely such that both the healthy development of AI technology and the protection of rightsholders and creators' rights are ensured." CODA's announcement states that OpenAI's Sora 2 AI video generator includes an opt-out system for copyright holders. However, under Japan's copyright laws, this after-the-fact approach to copyright infringement doesn't mean that OpenAI avoids liability. The organization's primary concern is the unauthorized use of its members' content for training purposes without their permission. "In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement," the organization explains. OpenAI's Sora 2 launched on September 30, 2025. In a blog post that went live within the first week of its launch, OpenAI CEO Sam Altman wrote that the plan was to "give rightsholders more granular control over generation of characters" and that for companies like Japanese game developers, Sora 2 represents a "new kind of engagement" that has a lot of value. In fact, even though the controls will allow rightsholders to exclude their characters and IP from Sora 2 videos, the goal is to "make it so compelling that many people want to." There's no denying that Sora 2 is impressive in its current state, but making it "compelling" for copyright holders doesn't exclude the fact that their data was used to train the model.

Square Enix, Bandai, and other Japanese studios demand OpenAI stop using their content without permission, drop a not-too-subtle hint about legal trouble if it doesn't

Japan's Content Overseas Distribution Association has noticed that an awful lot of what comes out of Sora 2 "closely resembles Japanese content or images." The Content Overseas Distribution Association, a Japanese industry group whose members include Square Enix, Bandai Namco, FromSoftware parent Kadokawa Corporation, and Cygames, has issued a formal notice to OpenAI demanding that it stop using its members content to train its Sora 2 video generation tool without permission. In a way, it feels like this was bound to happen. The day after Sora 2, the latest iteration of OpenAI's text-to-video technology, went live, PC Gamer's Jess Kinghorn described the system as capable of "spitting out all of the soulless, AI-generated Studio Ghibli-style animation one could ever want." Just a few days later, someone posted a Sora 2-generated video of OpenAI CEO Sam Altman standing in a field with frolicking Pokémon, saying, "I hope Nintendo doesn't sue us." All of this has attracted the attention of CODA, an organization founded in 2002 to combat piracy and promote Japanese videogames, film, music, animation, and television programs worldwide. Somewhat ironically under the circumstances, Nintendo is not a CODA member, although Studio Ghibli is. "CODA has confirmed that a large portion of content produced by Sora 2 closely resembles Japanese content or images," the org said in a written request to OpenAI (via Game Developer). "CODA has determined that this is the result of using Japanese content as machine learning data. In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement." No legal action has been taken at this point, but CODA gently hinted that it's not above such things if push comes to shove: "According to media reports, it has been stated that Sora 2 responds through an opt-out system based on requests from copyright holders. However, under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections." Notably, Altman promised after his romping with Pokémon video (and a bunch of others) rolled out that OpenAI will "give rightsholders more granular control over generation of characters [in Sora 2], similar to the opt-in model for likeness but with additional controls." Perhaps presaging the CODA demand, he also acknowledged "the remarkable creative output of Japan -- we are struck by how deep the connection between users and Japanese content is!" CODA's request to OpenAI contains two demands: That OpenAI ensure that CODA members' content isn't used for AI training without permission, and that OpenAI "responds sincerely to claims and inquiries from CODA member companies regarding copyright infringement related to Sora 2's outputs." I've reached out to OpenAI for comment and will update if I receive a reply.

Studio Ghibli politely declares war on Sora's AI slop

Just as one front in the battle between AI vs art closes, another opens. This week, the UK High Court ruled against Getty Images' claim that Stable Diffusion's training on images from its stock photo library constituted copyright infringement. But meanwhile, Studio Ghibli and other Japanese companies have finally had enough. The Japanese animation studio's distinctive art style has been one of those most ripped off using AI image generators. We've seen that in the viral Studio Ghibli-style memes made using OpenAI's Chat GPT 4, but it's OpenAI's Sora 2 video model that finally earned an incensed but very polite response. Will OpenAI CEO Sam Altman finally be shamed into changing his X profile pic at least? Studio Ghibli and other Japanese IP holders like Bandai Namco and Square Enix are represented by an anti-piracy body called the Content Overseas Distribution Association (CODA). It's sent a letter to OpenAI asking it to stop using its members' content for machine learning without permission and to responds sincerely to claims and inquiries from CODA member companies regarding copyright infringement related to Sora 2's outputs. "CODA, in cooperation with its member companies, requests that OpenAI respond sincerely such that both the healthy development of AI technology and the protection of rightsholders and creators' rights are ensured," the letter states. The complaint isn't that Sora produces video that resembles its members' IPs, but that OpenAI used those IPs for training without permission. OpenAI had initially said that copyright owners would have to opt out if they didn't want Sora to use their art to create AI deepfakes. It expressed surprise when a wide range of rights holders turned out to have a problem with that and has since backtracked. It now claims to block the generation of copyright IPs unless their owners have opted in. But it's not clear if that's a good enough response for CODA as copyright work may have already been used in the training. "Under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections," the letter states. Japan's government has also asked OpenAI not to replicate Japanese artwork. Studio Ghibli co-founder Hayao Miyazaki hasn't commented on the viral explosion of AI imagery in the studio's style but he famous described AI as "an insult to life itself" when he was shown an early experiment in AI-generated 3D animation in almost a decade ago.

Japanese Studios Ask OpenAI to Stop Using Anime for AI Training

A major Japanese trade group representing publishers and studios, including Studio Ghibli, has asked OpenAI to stop using their copyrighted content to train its AI models. The appeal, made by the Content Overseas Distribution Association (CODA), follows growing concerns that OpenAI's tools, especially its new video generator Sora 2, may have learnt from Japanese media without permission. This latest move follows earlier action by the Japanese government, which had also requested OpenAI in early October to refrain from infringing on the copyright of Japanese cultural products. Minoru Kiuchi, Japan's Minister of State for Intellectual Property and AI Strategy, revealed on October 10 that the government's Intellectual Property Strategy Promotion Secretariat had made an official online request to OpenAI after anime-style videos using copyrighted characters from Sora 2 surfaced on social media. In its recent letter, CODA stated that "a large portion of content produced by Sora 2 closely resembles Japanese content or images," suggesting that OpenAI trained its model using Japanese works. The association warned that reproducing copyrighted material during machine learning "may constitute copyright infringement." CODA also objected to OpenAI's opt-out policy, which allows rights holders to request removal of their works after training. The group stressed that Japan's copyright system requires prior permission for such use. "Under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections," CODA said in its letter. The organisation has asked OpenAI to ensure that its members' works are not used for training without consent and to respond "sincerely" to any infringement complaints. CODA said it aims to protect creators' rights while supporting responsible AI development. Concerns about copyright violations intensified soon after Sora 2's launch on September 30, when users began sharing anime-style videos resembling well-known characters from Japanese films and series. Japanese MP Akihisa Shiozaki wrote on X (formerly Twitter) on October 5 that the reproduction of copyrighted anime characters was "a very serious issue" that could "threaten the Japanese creative industry." OpenAI CEO Sam Altman addressed the issue on October 4, announcing new measures to give rights holders "more granular control" over how their characters could be used in AI-generated content. Altman also said OpenAI planned to monetise Sora's video generation feature and share part of the revenue with rights holders, a move that some critics see as an attempt to commercialise copyrighted content before clear consent mechanisms are in place. Japan's position highlights a complex contradiction in its AI policy. The country has previously aimed to be "the world's most AI-friendly nation." A 2024 government white paper allowed AI developers to train models on copyrighted content, even for commercial purposes. However, the Japan Copyright Office clarified that it considers AI-generated content infringing only when it shows "similarity or dependence" on an existing copyrighted work, while merely imitating a "style" does not violate the law. This legal grey area has complicated enforcement, especially as OpenAI's models can generate near-exact reproductions of Japanese animation aesthetics. The issue also mirrors ongoing global litigation, including Getty Images v. Stability AI, The New York Times v. OpenAI, and ANI v. OpenAI in India. In June 2025, a U.S. District Judge in Bartz v. Anthropic ruled that training on purchased copyrighted works falls under fair use, but training on pirated data does not. Anthropic later settled for $1.5 billion, marking the first major win for rights holders in an AI-related copyright case. Japan's animation industry, valued at $22 billion in 2023, is one of the country's most influential global exports. With AI-generated videos and images now blurring the line between inspiration and imitation, authorities are keen to protect this key sector. While OpenAI's opt-out model aims to balance innovation and copyright, Japanese lawmakers and creators argue that post-facto exclusion does not meet the country's legal standards. The government's recent interventions indicate that Japan may tighten copyright enforcement for AI-generated content despite its otherwise liberal stance on AI training. Studio Ghibli has not issued an official response, even though AI models frequently mimic its works. But co-founder Hayao Miyazaki's earlier criticism of AI-generated animation remains widely cited. When shown an AI demo in 2016, he said he was "utterly disgusted", adding, "I can't watch this stuff and find it interesting. I feel strongly that this is an insult to life itself." The dispute between CODA, the Japanese government, and OpenAI underscores a broader global question: can AI progress without undermining creators' rights? Japan's dual approach, promoting AI growth while protecting cultural heritage, illustrates both the promise and the peril of generative technologies. As AI-generated art grows more sophisticated, how Japan reconciles innovation with intellectual property protection may shape global copyright norms in the age of artificial intelligence.

Studio Ghibli and other Japanese publishers demand OpenAI to stop using their content for AI training

The dispute follows AI tools mimicking Ghibli's animation style. A Japanese trade group representing publishers and studios, including Studio Ghibli, has asked OpenAI to stop using their copyrighted materials to train artificial intelligence models without permission. The request came in a letter sent last week by Japan's Content Overseas Distribution Association (CODA). Studio Ghibli, famous for films like Spirited Away and My Neighbor Totoro, has been particularly affected by the spread of AI-generated images and videos mimicking its animation style. When OpenAI's image generator was released in March, many users began creating Ghibli-style versions of themselves, their pets and other photos. Even OpenAI CEO Sam Altman joined in by changing his profile picture on X to a Ghiblified image. With OpenAI's video tool Sora now becoming more widely available, CODA has urged the company to stop using its members' creative works in training data without permission, reports TechCrunch. Also read: Apple iOS 26.1 update starts rolling out: What's new, how to download and supported devices "In cases, as with Sora 2, where specific copyrighted works are reproduced or similarly generated as outputs, CODA considers that the act of replication during the machine learning process may constitute copyright infringement," CODA wrote in the letter. "Under Japan's copyright system, prior permission is generally required for the use of copyrighted works, and there is no system allowing one to avoid liability for infringement through subsequent objections." Also read: OpenAI and Amazon sign $38 bn deal for ChatGPT-maker to use AWS: Here's how the partnership works Meanwhile, US law remains unclear on whether training AI with copyrighted material is illegal. A recent court ruling found that another AI company, Anthropic, did not break copyright law by training on protected books, though it was fined for pirating them. Also read: Samsung Galaxy Z Flip 6 available with over Rs 43,500 on Amazon: How to grab this deal