ChatGPT Ghibli Trend Weaponizes ‘Harmlessness’ to Manufacture Consent

In late March, OpenAI CEO Sam Altman announced that the company’s flagship AI platform, ChatGPT, could produce high-quality images under its new version GPT-4o. Within hours, a Seattle software engineer used the new capabilities to transform a family photo into the style of Studio Ghibli films like Howl’s Moving Castle, Princess Mononoke, and Spirited Away. The trend quickly went viral, with Altman and other OpenAI staffers posting Ghiblified images.

As users overloaded OpenAI servers conjuring up these images, the platform gained a million new users within an hour, eventually pushing the platform past 150 million users. By March 27, the White House, the Israel Defense Forces, and India’s civic engagement platform had all jumped on the trend to post Ghiblified propaganda on X. But all users posting Ghibli images—whether they realized it or not—were participating in propaganda.

Related Articles

While the term propaganda tends to conjure images of George Orwell’s 1984 or the graphic posters of Hitler’s Third Reich, propaganda in the 21st century operates differently. As political scientist Dmitry Chernobrov wrote recently, in the age of social media, “the public themselves are co-producing and spreading the message [of elites], meaning that manipulative intent is less evident, and the original source is often obscured.” Similarly, artist Jonas Staal wrote in Propaganda Art in the 21st Century (2019) that propaganda’s basic aim is a “process of shaping a new normative reality that serves the interests of elite power” regardless of whether its deployed in a democracy, a dictatorship, or by a corporation.

It might seem harmless to transform a family photo into the cutesy animated aesthetic of a Hayao Miyazaki film. And yet it reads as more sinister when the White House posts a Ghiblified image of an immigrant without legal status being deported. But, if both images are a kind of messaging to manufacture consent, we have to ask: consent for what exactly?

Since AI image generators were released to the public in 2022, OpenAI and competitors Midjourney and Stability AI have faced pushback. It is widely understood that these models were trained on billions of images scraped from the internet—without creators’ consent or compensation—a fact that has led artists to sue the companies for copyright infringement. Thus far no such lawsuit has been successful. (An OpenAI whistleblower set to testify about IP violations was found dead late last year—his parents are investigating claims it wasn’t a simple suicide.) Aside from court cases, artists have fought a battle for public opinion, urging users and companies to boycott the use of AI.

The fact that the Ghibli images became the symbol of the rollout—as opposed to bizarre “AI slop”—wasn’t mere luck. Altman, as the very public face of his company’s brand, made sure to participate in the trend, changing his profile picture to a Ghiblified portrait and reposting the cute and seemingly legitimate images made with the Ghibli edit. In fact it was the only imagery he used on his social media to promote the update.

Meanwhile, Miyazaki’s work is so well known, and GPT-4o’s mimicry so convincing, that the launch resembled a real brand partnership with one of the world’s most respected animation studios. (That OpenAI managed to rip off a well-known brand and get away with it is an advertisement in itself). While Studio Ghibli still hasn’t commented, considering Miyazaki once called AI art “an insult to life itself,” it’s safe to assume he did not consent to his work being used in this way. 

But it isn’t just Altman and the OpenAI team who are defanging AI technology—it’s every user who participates. Every time a user transforms a selfie, family photo or cat pic into a Ghibli-esque image, they normalize the ability of AI to steal aesthetics, associations, and affinities that artists spend a lifetime building. Participatory propaganda doesn’t require users to understand the philosophical debates about artists’ rights, or legal ones around copyright and consent. Simply by participating they help Altman and his competitors win the battle of public opinion. For many, the Ghibli images will be their first contact with generative AI. For more informed users, the flood of images reinforces the inevitability that AI will re-shape our world. It’s particularly cruel that Miyazaki’s style has become the AI vanguard, given that he famously stuck to laborious hand-drawn animation even as the industry shifted to computer-generated animation.

The leveraging of Ghibli images is not limited to Altman’s AI evangelism. Far more sinister has been the use of Miyazaki’s aesthetic towards ends that directly contradict his long held anti-war and anti-fascist politics.

Growing up in World War II-era Japan and its aftermath, Miyazaki embedded anti-war themes throughout his cinematic oeuvre. When he won an Oscar for Spirited Away in 2003, he refused to attend the ceremony in protest of the Iraq War. Howl’s Moving Castle (2004) and Porco Rosso (1992) depict characters who would rather live on the run than work for imperialist warmongers. Studio Ghibli’s hand-drawn style cannot be separated from the films’ humanist politics. Miyazaki and his team designed a rich and versatile aesthetic housing themes of environmentalism, resistance to war, and an ethic of care. His films are driven by characters who become brave, responsible, and moral in response to shows of love and solidarity. Miyazaki invites us into a world where strength of character is just as wondrous as every epic flight or flash of magic.

Studio Ghibli spent decades developing these humanist associations. But, in a matter of days, political actors hollowed out his aesthetic. On March 27, the official account of the White House posted an image of Virginia Basora-Gonzalez—previously convicted for fentanyl trafficking and lacking legal status—being handcuffed next to an American flag. The image wasn’t a photograph, but a cartoon rendered in the Ghibli aesthetic.

(It’s not the first time Trump’s team used generative AI to make a tasteless and cruel point. In February, Trump posted an AI-generated video showing Gaza under Trump’s control. Ironically, the bizarre video—which showed bearded dancers in bikinis and a giant Trump statue—was originally created as satire, according to the video’s creators.)

Ghibli-fying Basora-Gonzalez was calculated: she was unlikely to draw sympathy compared to, say, Kilmar Armando Abrego Garcia, Lewelyn Dixon, and Camila Muñoz—all since-deported immigrants with legal status and American families. By rendering Basora-Gonzalez into a cartoon, they turn her, and her plight, into a stereotype and a caricature. It’s easier to dismiss a cartoon crying, than a real woman in distress. It also primes regular people to mock the plight of other brown people caught up in Trump’s immigration crackdown. When all the specificity of the photograph is erased, Basora-Gonzalez is made into just one more Latina service worker, inviting audiences to read this stereotyped character as always potentially guilty and deserving of deportation–perhaps not for dealing fentanyl but any crime. Like lacking documents. Or being an American child of undocumented parents. Or having the wrong tattoo.  Or – maybe Latinos should be deported for no reason at all. Notably, the border patrol agent bears little resemblance to those who arrested Basora-Gonzalez: instead of a masked figure in body armor, the man appears older, stern, even sympathetic in an olive green quarter-zip. He just looks like a Boomer-age man you’d see around town.

If GPT-4o and future models help nefarious governments, OpenAI’s Altman doesn’t seem to mind. Like Meta CEO Mark Zuckerberg before him, he casts OpenAI as the innocent builder of a neutral platform that, in the name of free speech, might be used by bad actors. In a post on X, Altman defended GPT-4o’s ability to generate “offensive stuff,” writing, “we think putting this intellectual freedom and control in the hands of users is the right thing to do.” By addressing debates about moderation and freedom of speech on social media, but not intellectual property theft, Altman positions users as political, not technology. So when governments weaponize AI for vulgar propaganda, it doesn’t necessarily damage the aura of harmlessness Altman’s team is clearly trying to build around OpenAI. It keeps the conversation focused on the content of these images, not whether the tools should exist at all. 

In The Work of Art in the Age of Mechanical Reproduction, Walter Benjamin famously warned, upon seeing film’s rising popularity, that technologies enabling mass reproduction of images could fuel fascism. He reasoned, “Fascism sees its salvation in giving the masses not their right, but instead a chance to express themselves.” Social media and generative AI similarly offer frictionless expression, from re-sharing or participating in a viral trend to commenting, that masks their disempowerment as political or cultural agents. This expression is obviously powerful: it’s hard to see Trump’s political rise, Elon Musk’s centrality to American culture, or today’s memeified politics without it.

Generative AI, Altman insists, has the power to change the very nature of our social contract. But, more immediately, it represents an increasing appropriation of art by Big Tech. One is left to wonder, as the technology appears to wipe out the value of artistic skill and intellectual labor with each update, “What is art for?”

In an essay for the New Yorker, sci-fi author Ted Chiang punctured AI’s triumphalism with a simple truth. The point of art, he wrote, isn’t the final product of expression, nor is selling or showing work the point of expression. It is the process of struggling to write a perfect sentence or to hand-draw an animated film with 40 other artists, as Miyazaki did, that makes meaning. True expression requires committing yourself to seeing something through in its most awful, embarrassing stages, to withstand critique and doubt, to sustain feelings of discomfort and vulnerability not just for a few moments but for weeks, months, years, even a lifetime.

As Miyazaki has conveyed so compellingly in his films, it is laborious, loving, useful work that imparts self-knowledge and the understanding of shared struggle and humanity. You cannot skip to the end. You cannot just generate the artwork or the essay and get the learning and the satisfaction that the work imparts. When you forgo that labor, you let fascists, corporations, and all manner of elite actors make the act of expressing yourself so easy that you feel free, even agentive, as they take your rights from you.