A new obsession has taken up social media flows: Ghibli- style portraits. Elon Musk’s Openai and Grok 3 chatgpt gave users the opportunity to transform their selfies into dreamlike scenes that recall the cinematographic masterpieces of Japanese host Hayao Miyazaki. At first glance, it is a harmless pleasure. Who would not want to be redesigned with the sweet fantasy of cult films Fiery Or My neighbor Totoro? But hiding under nostalgia and the aesthetic delight is a tacit question: what is happening to all these faces?
The jar of honey of biometric data
The launch of the Ghibli-d’Openai style generator last week sparked an instant viral storm. Politicians, celebrities and everyday users rushed to download their photos, eager to be reinterpreted through the signing brush strokes of the Ghibli studio. Grok 3, the Musk Chatbot AI, was not far behind, adding its own Ghibli-de style ghibli-aux offers from Xai. And just like that, thousands, maybe millions of people, of people have gladly given new facial data to AI platforms.
Privacy supervisors have taken note. On X (formerly Twitter), digital rights activists agitate red flags, warning that these AI generators could be Trojan horses designed to raise a large database of human faces. Unlike web data, which fall under the regulatory examination, these downloads are voluntarily subject, avoiding legal roadblocks as the “legitimate interest” test of the GDPR. According to Luiza Jarovsky, co -founder of the AI, Tech & Privacy Academy, this transforms the collection of OpenAi data in a different legal category – which allows greater freedom in the processing of biometric information.
🚨 Most people have not realized that the Ghibli effect is not only a controversy on the copyright of the AI, but also the Trart of Openai public relations to have access to thousands of new personal images; Here’s how:
To get their own Ghibli version (or Sesame Street), thousands of people are now voluntarily… pic.twitter.com/zbktsy
– Luiza Jarovsky (@luizajarovsky) March 29, 2025
“People give OPENAI to know a gold mine of fresh and high quality facial images to OPENAI, wrote Jarovsky in an article. “Their policy explicitly stipulates that user inputs can be used to form AI unless they undress.”
This raises a frightening perspective: Openai and Xai now have access to a flood of images provided by the user, many of which include clear and high resolution faces: ideal material to refine facial recognition and Deepfake technologies. Even if Openai and Xai claim not to store these images, what insurance do users have?
Has the trend been designed?
Some skeptics believe that Ghibli AI’s trend has not become organically viral. An increasing theory suggests that the CEO of Openai, Sam Altman, deliberately enlisted celebrities and influencers to first publish their portraits generated by AI, ensuring that the trend quickly spread to regular users. By creating FOMO (fear of missing), the company could accelerate mass adoption and, in turn, collect an unprecedented volume of facial data in record time.
It would not be the first time that a technological company has exploited influencer marketing to push new feature, but if it is true, it raises ethical questions. Was it a calculated decision to build training data on AI under the cover of a viral trend? What if celebrities were there, have they disclosed promotional agreements? Openai did not comment on these speculations, but the coordinated nature of the trend has left many questions its origins.
The silent cry of the Ghibli studio
The ethical mine extends beyond privacy. For years, the Japanese host Miyazaki, now 84, has made his disdain for AI and clear automation. In 2016, a video of him reacting to an animation generated by AI became viral … His disgust was palpable. “I strongly think that it is an insult to life itself,” he said.
That the models of AI now generate ghibal -style images using its aesthetics without its consent would probably cause a similar answer. The art generated by AI which imitates human creativity is already at the center of an intense copy of copyright, and this latest trend adds another layer of discomfort: do the tools of AI actively erod the heritage of master artists while collecting user data?
Chatgpt response
While Openai has not yet published an official data security declaration, Chatgpt’s own response to a user’s request was revealing:
“No, it is not certain to download personal photos on any AI tool, unless you are certain of its privacy policies and its data processing practices. OPENAI does not keep or uses downloaded images beyond immediate session, but it is always better to avoid sharing sensitive or personal images with AI services.”
Grok 3 was more elusive on the quick question:
“XAI does not explicitly detail the duration of the downloaded images or if they are used to form future models. Many AI tools store downloaded data on servers, where they could be vulnerable to violations. Unless you withdraw (check the X settings if you access Grok via the platform), your images can also feed the improvement in AI.”
So, if a user does not take proactive measures to withdraw, there is a high possibility that his images will be stored and reused.
Biometric arms race
Risk is not only a question of AI training. Biometric data is among the most precious digital products. Cybersecurity experts warn that these downloaded images could be used in a way that users have not envisaged. Once an image is processed, AI systems theoretically have the ability to generate new variations in a person’s resemblance, which raises concerns about identity theft and improper use.
Himachal Cyber ​​Warriors, a cybersecurity collective, has issued a striking warning: “Think before you #Ghibli. This cute ghibli-style selfie selfie ghibual style? It could cost you more than you think.
How to protect your privacy
As the art generated by AI-AI becomes more and more sophisticated, users must approach these trends with caution. Before downloading a selfie to the next viral AI tool, consider these confidentiality guarantees:
- Think twice before sharing personal photos … in particular high resolution images.
- Use anonymous or artistic avatars instead of real selfies.
- Review the confidentiality policies of the AI ​​platforms and remove data retention as much as possible.
- Limit authorizations for applications that require access to the camera or gallery.
- Avoid facial recognition for unlocking devices. Use a pin instead.
Ghibli’s controversy underlines a critical point: AI’s progress exceeds public awareness of their implications. Technological companies like Openai and Xai explain these features as innocent and entertaining, but they are often delivered with invisible compromises.