Can a machine mimic a masterpiece without breaking the law? That question is at the heart of a growing debate sparked by OpenAI’s new “Ghiblifying” feature, which allows users to generate artwork that echoes the signature style of Studio Ghibli’s animation.
The feature, rolled out with the March 25 release of GPT-4o, transforms user-submitted images into soft watercolor renderings reminiscent of Ghibli classics like Spirited Away and My Neighbor Totoro. The capability has gone viral, driving a spike in user engagement that helped push ChatGPT’s weekly active user base to more than 150 million, according to Similarweb. Yet amid the enthusiasm lies an unresolved conflict over copyright, transparency, and artistic authenticity.
U.S. copyright law protects specific creative expressions but not artistic styles in the abstract. That distinction leaves AI-generated content like Ghiblified images in a legal gray area. Critics argue that while the tool may not replicate specific Ghibli scenes or characters, it leverages a look so distinctive it blurs the line between homage and imitation. Legal experts warn that if OpenAI’s model was trained on copyrighted Ghibli content without authorization, it could expose the company to infringement claims—though proving such training data use is nearly impossible without full transparency.
Studio Ghibli co-founder Hayao Miyazaki, a vocal critic of AI in art, has long lamented the role of technology in diminishing creative effort. Although his past comments predate GPT-4o, his stance echoes widely held sentiments among traditional artists who see generative AI as a threat to originality and labor-intensive craft.
OpenAI maintains it is taking steps to avoid infringing on copyrighted material. GPT-4o blocks attempts to generate images in the style of specific artists or copyrighted properties and instead offers broader descriptors like “anime watercolor.” The company has positioned these safeguards as part of a balanced approach to foster creativity while respecting intellectual property rights. However, skeptics argue that generalized style imitation, particularly when so closely associated with a single studio’s aesthetic, remains ethically suspect and potentially damaging to creative industries.
The viral spread of Ghiblified content across social media platforms has only intensified scrutiny. Analytics from SensorTower show a marked increase in app downloads following the feature’s launch, overwhelming OpenAI’s infrastructure and prompting the implementation of rate limits to stabilize usage.
Under current U.S. law, AI-generated works cannot be copyrighted unless they include meaningful human authorship. That ruling by the U.S. Copyright Office further muddies the waters for outputs like Ghiblified art: while not protectable in their own right, they could still infringe on existing works if those were used in training the model. Japan’s copyright framework introduces additional complexity. While similar to U.S. law in emphasizing expression over style, Japanese law also grants creators moral rights—including the right to preserve a work’s integrity—which could be cited in potential legal disputes.
One of the core challenges for rights holders lies in the lack of transparency around training data. OpenAI, like most major AI developers, has not disclosed the full contents of its training datasets. This opacity makes it nearly impossible to determine whether proprietary material was used and undermines any legal claim that requires proof of unauthorized input. The gap has prompted growing calls from policymakers and IP advocates for mandatory dataset audits or documentation requirements.
In tandem, the ethical debate continues to intensify. Many artists and commentators argue that generative tools risk cheapening the value of human-created art by reducing distinctive styles to algorithmic filters. Some have proposed requiring artist consent before AI models can emulate identifiable aesthetics. Others suggest that AI should serve a complementary role—handling repetitive tasks while leaving higher-order creativity to humans.
Regulators are beginning to respond to these concerns. The European Union’s upcoming AI Act includes transparency mandates and rights-holder notification clauses. In the United States, lawmakers have held congressional hearings on AI and intellectual property, though comprehensive legislation has yet to emerge.
As the capabilities of generative AI evolve, so too do the risks and responsibilities for developers, artists, and policymakers alike. The debate over “Ghiblifying” is not just about one company or one studio—it’s a bellwether for how societies will manage the balance between technological progress and creative integrity.
And so, the central question remains: can a machine mimic a masterpiece without crossing legal or ethical lines? As OpenAI’s Ghiblifying feature continues to spread, the answer may shape the future of not only AI, but art itself.
News Sources
- Ghibli effect: ChatGPT usage hits record after rollout of viral feature (Yahoo! Finance)
- The Ghibli effect: ChatGPT’s AI art craze drives record-breaking growth (Ctech)
- Studio Ghibli Has Few Legal Moves Against OpenAI ChatGPT Image Feature (Business Insider)
- OpenAI responds to criticism of ChatGPT’s Studio Ghibli-style images (Mashable)
- After Studio Ghibli AI Craze, Sam Altman Teases Powerful ‘Images V2’ Update (Live India)
Assisted by GAI and LLM Technologies
Source: HaystackID published with permission from ComplexDiscovery OÜ
