By By Nathalie DEVILLIER, Ph.D. in International Lawl
This case serves as a reminder that it is essential to look beyond the media hype, especially when it may threaten our fundamental rights, including the right to privacy.
Over the past few days, images generated by OpenAI’s latest tool have been flooding social media. These creations, heavily inspired by the style of Studio Ghibli—famous for films like My Neighbor Totoro and Spirited Away—have sparked a viral phenomenon dubbed the “Ghibli effect.” While this trend appeals for its aesthetic appeal, it raises crucial questions about copyright and the privacy of personal data.
When AI replicates copyrighted styles
From a legal standpoint, simply replicating an artistic style does not, in and of itself, constitute copyright infringement. However, the ability of AI to generate images in a specific style with uncanny accuracy raises a major question: was it trained on copyrighted works?
Has OpenAI obtained licenses from Studio Ghibli or other creators to use their works as training data? There is no evidence to support this claim. In the United States and elsewhere, lawsuits are mounting against OpenAI and Anthropic for copyright infringement. Authors, artists, and creators are denouncing the unauthorized use of their works by AI models, without compensation or prior agreement.
This debate goes far beyond the realm of art alone. If AI can recreate images in a protected style, where does the line between legitimate inspiration and unauthorized reproduction lie?
Personal images in the public domain: the trap of implied consent
The “Ghibli effect” is not just a copyright issue; it also has major implications for the protection of personal data.
Within the European Union, OpenAI justifies training its models on images collected online by invoking “legitimate interests” (Article 6(1)(f) of the GDPR). However, this legal basis imposes strict obligations, including the implementation of safeguards to prevent the misuse of sensitive data.
But another issue arises: user consent. To get a Ghibli-style portrait of themselves or one in another style, thousands of internet users voluntarily upload their photos to the platform. In doing so, however, they accept OpenAI’s terms and conditions and authorize the company to use their images (Article 6.1.a of the GDPR).
The result? OpenAI gains free access to a massive database of new faces without having to extract them from social media. Worse still, these photos may include private images—such as family photos or intimate pictures—that were not originally available online. Meanwhile, competing platforms can only see the “Ghibli-fied” version of these images, giving OpenAI a significant strategic advantage.
Toward stricter regulation of generative AI
The consequences of this situation are far from trivial. On the one hand, artists find themselves up against a system capable of replicating their work without giving them a say in the matter. On the other hand, users unwittingly provide the platform with a valuable resource: their personal data.
In light of these challenges, regulators around the world are beginning to take action. The European Data Protection Board reiterated in its Opinion 28/2024 that these platforms must strictly comply with the principles of the GDPR. Other jurisdictions, such as certain U.S. states, are considering stricter regulations to govern the use of personal data by AI.
But will these efforts be enough? Artificial intelligence is evolving faster than legal frameworks, and the risks of misuse persist. Until clear and effective safeguards are put in place, this legal vacuum will be exploited by companies at the expense of artists and users’ privacy.
The “Ghibli effect” highlights the growing tensions between artificial intelligence, copyright, and data privacy. On the one hand, artists are seeing their styles replicated without any control. On the other, internet users are unwittingly handing over sensitive information to an AI that is becoming increasingly insatiable.
These issues are not trivial: they shape the future of creativity and privacy in the age of AI. Far from being merely an aesthetic trend, this phenomenon highlights the major challenges that lawmakers and regulators will face in protecting the rights of creators and citizens in the face of new technological advances.
