Upon its release last month, OpenAI’s GPT-4o, which allowed people to transform any picture into the artistic style of Studio Ghibli, became a global sensation. The bandwagon was joined by folks from all walks of life and in no time, there were Studio Ghibli versions of U.S. President Donald Trump, Elon Musk, and others all over the internet.
The rush caused a meltdown of GPUs, forcing OpenAI CEO Sam Altman to announce user limits on image generation.
A previous attempt at image models made by Google’s Gemini 2.0 Flash sparked controversy when it was used to remove watermarks from copyrighted images. The current wave rekindles the debate around the issue of whether training AI models on copyrighted works constitutes fair use, and what security artists have amidst such technological advances.
The primary concern revolves around whether AI-generated images that replicate the style of Studio Ghibli violate copyright laws. According to IP lawyer Evan Brown, while the specific style of an artist is not explicitly protected by copyright, the use of copyrighted works to train AI models poses a legal dilemma. If AI systems are trained on copyrighted materials without permission, this could constitute a violation of copyright law.
However, the existing copyright laws do not adequately address the complexities introduced by AI. For instance, debate centers on whether the visual style of a living artist can be copyrighted. As of now, there are no clear legal precedents that definitively resolve these issues, leading to a situation where companies like OpenAI are operating in a “legal grey area” regarding the generation of art in the style of living artist.
Moreover, the situation is not unique to Studio Ghibli. There have been other instances. For example, OpenAI is facing lawsuits from various creators, including the New York Times, for allegedly using their content without permission to train its models. Likewise, there have been cases where AI-generated images closely resembling popular anime characters have led to legal action, highlighting the ongoing struggle between innovation in AI and the protection of creative works.
Hayao Miyazaki, co-founder of Studio Ghibli, has long been vocal about his disdain for AI-generated art. He has described it as “an insult to life itself,” expressing concerns about its impact on creativity and authenticity. Miyazaki’s criticism stems from a belief that AI lacks the emotional depth and human experience that characterize true artistic expression. His views reflect a broader apprehension among artists regarding the potential devaluation of human creativity in the face of advancing technology.
It also disregards the time and labor a single frame of animation takes to come to life. For instance, Miyazaki was adamant that the scene of a busy market for the movie “The Wind Rises” (2013) had to be hand drawn, without using any CGI. It took animator Eiji Yamamori one year and three months to complete the four-second scene. At 24 frames per second, a total of 96 images, resulting in 6.4 images per month. AI allows people to bring this effort and time to negligible level, and takes away the essentials of craft, along with putting the originality of artists at stake.
The challenges posed by AI-generated art necessitate a multifaceted approach that includes revising existing laws, implementing safeguards, and encouraging ethical practices.
The most important step is to revise and update copyright laws. Current copyright laws were established before the rise of AI technologies and do not adequately address the nuances of AI-generated content. Lawmakers must establish clear definitions of what constitutes an “original work” in the context of AI. This includes defining the extent of human involvement required for copyright eligibility. For instance, the U.S. Copyright Office has ruled that works generated solely by AI lack the necessary human authorship for copyright protection.
In addition, guidelines that govern how AI models can be trained must be created. This could involve requiring explicit consent from copyright holders before their works are used in training datasets. For example, like the European Union’s Copyright Directive, which emphasizes the need for transparency and fair compensation for creators, AI laws could adopt similar measures to protect artists’ rights.
Furthermore, there is a need to develop legal frameworks that recognize hybrid authorship, where both AI and human contributions are acknowledged. This could involve establishing a system where the human creator retains rights over the final output, even if AI played a significant role in its creation.
To prevent unauthorized use of artistic styles, companies developing AI technologies must implement safeguards. AI platforms like OpenAI have begun implementing content filters that prevent the generation of images in the style of living artists. These filters should be continuously updated to adapt to new artistic styles and emerging artists. Similarly, User Agreements can be introduced, requiring users of AI art generators to agree to terms that prohibit the use of copyrighted styles without permission. This could work the way music streaming services manage licensing agreements with artists and record labels. Companies should also establish monitoring systems that allow artists to report unauthorized use of their styles or works. This could involve partnerships with copyright organizations to facilitate swift action against infringing content.
In addition, the discourse around the ethical use of AI needs to be strengthened. The creative community should actively promote ethical practices in AI art generation. Collaboration between artists and AI developers must be promoted to create frameworks that respect artistic integrity while fostering innovation. For instance, initiatives like the “Creative Commons” allow artists to share their works under licenses that specify how others can use them, which could be adapted for AI-generated content.
Moreover, educational programs that inform artists about their rights and the implications of AI in their work should be developed. This could include workshops on how to protect their art from unauthorized use and how to navigate the evolving landscape of copyright law.
Lastly, establishing ethical guidelines for AI developers that emphasize respect for original creators. This could involve creating a code of conduct that AI companies must adhere to, like the ethical standards set by professional organizations in other fields, such as journalism or medicine.
International copyright frameworks, like the European Union Copyright Directive, can provide valuable insights. The EU has implemented measures that require platforms to negotiate licenses with rights holders for the use of their works. This model could inspire similar legislation in the field of AI to ensure that artists are compensated for the use of their styles in AI training.
Other such models include the U.S. Fair Use Doctrine which allows for limited use of copyrighted material without permission under certain circumstances. This principle could be expanded to include provisions specifically addressing AI-generated content, ensuring that artists can still benefit from their work while allowing for innovation in AI.
While the issue around Art and AI would continue to evolve based on the continuous upgrading of the technology, the best way to preserve the peculiarities of artists while allowing innovation to flow free is by bringing the subject under rightful laws and ethical standards, which would serve as the standard of operation and evolution for technologies to come.