By: Kelsey Shaughnessy
World renowned art auction house, Christie’s, proudly announced its inaugural Artificial Intelligence (“AI”) art auction, Augmented Intelligence, in February. The groundbreaking auction was dedicated to art created or enhanced with generative AI tools, like DALL·E, Midjourney, and Stable Diffusion. On view for only a short period, from February 20 to March 5, the auction was visible both online and in person at Christie’s Rockefeller Center galleries in New York City. The auction contained thirty-four works and generated $728,784 of sales. Christie’s considered the auction to be a natural progression in art history and reverently adopted AI art as being an art form that is “undoubtedly the future.” However, the auction was met with fierce opposition. Thousands of artists wrote a letter demanding Christie’s cancel Augmented Intelligence, explaining grave concerns over the artwork featured in the auction and how these works were created by generative AI models trained on unlicensed copyrighted artwork. Given two divergent opinions, artists and those who sell art, both toil with the same question: as technology evolves, how do artists collaborate with technology? The answer likely lies in the refuge of intellectual property law and judicial decisions from pivotal generative AI cases.
Throughout history, artistic movements have addressed cultural zeitgeists head-on, reflecting society’s mindset and values. AI is no different, following a disruptive ideology commonly found in art history. For example, with the advent of photography in the nineteenth century, some artists felt embracing photography as an art form was a “banalization of artistic creation,” functioning only as an imitation of art with commercial utility, never able to fully develop into a true art form. Whereas other artists positively embraced photography as a new medium for expression, seen as a “democratization of art and image,” allowing art to become more affordable and accessible for the first time. As more artists opted for a camera as a medium, art moved away from realism, or exact portrayals of reality, moving toward Impressionism, starting the modern art movement. Artists of the nineteenth century were vanguards, exploring how to utilize new technology while wrestling with the ethics of its use as a new medium.
Similar to how artists and creators felt about photography, AI challenges traditional notions of art being technical, skillful, emotive, and beautiful. AI works by utilizing computers to simulate the human experience— including learning, comprehension, problem solving, decision making, even creativity and autonomy. Generative AI, such as ChatGPT, are AI models “trained” on text, images, and audio, to understand the relationships between words and concepts, drastically changing the way artists approach creativity. The process is equated to how an artist traditionally learned how to paint, studying and imitating the Old Masters. Generative AI models function by “memorizing” parts of the works they train on to be able to generate near verbatim reproductions of the works and produce “synthetic” search results in the form of verbatim text or images as a search result, effectively allowing users, like artists, to only utilize, e.g., ChatGPT, as their source, instead of the artwork itself.
Artists utilize generative AI to complete various creative tasks, whether wielding a brush, translating text-to-images, even sculpting three-dimensional models. Described as inpainting and outpainting, these techniques allow for artists to enable changes on existing images by simply uploading the image to the generative AI model. Once uploaded, minor alterations, i.e., changing shadowing, reflections, and textures, or major changes, i.e., extending the original image beyond the existing plane to create an entire new background, can be made on the image. Artists choose to upload their own images or use existing images to train their AI models, ultimately yielding their final work product.
Disrupting cultural and society norms, AI has also upended legal norms. Courts in the United States have sought refuge within intellectual property law. Copyright is a form of intellectual property protecting original works of authorship. Originating from the Constitution, an author has an “exclusive [r]ight” to their created works, to promote progress of Science and Arts. By registering an original work with the U.S. Copyright Office, a creator is granted legal protection against copyright infringement, or nonconsensual use, e.g., reproduction, distribution, display, or performance, of a registered creative work. With generative AI use in creative spaces, questions of authorship, data ownership, and ethical use, aggravate the use of AI in art.
With an appetite for avant-garde, Christie’s announced Augmented Intelligence with cautious language steering the narrative away from legal and ethical use of AI in the art. Christie’s described AI art as “any form of art that has been created or enhanced with AI tools,” focusing on collaboration between the artist and AI, and differentiating it is not a “shortcut to productivity [or] artistry . . . .” Over six thousand artists expressed concern of AI-generated art at the auction. Artists’ concerns were over authorship, data ownership, copyright infringement, and the loss of the human aspect of creativity. Moreover, how these ideals are encouraged and perpetuated if a prestigious art auction house like Christie’s decides to promote and sell art created by AI.
Augmented Intelligence raises continuing legal issues which need to be addressed to stabilize a healthy, sustainable relationship with artists and the use of AI. AI’s ability to wield a brush, translate text to images, and sculpt, is a phenomenon, allowing for artists to create unique works as disruptive as the technology utilized. This ideology resonates in the art world and Christie’s is leveraging it.
However, when generative AI models scrape, download, and process creative works, these models may be infringing on the right of reproduction protected by copyright law. Artists who requested Augmented Intelligence to be cancelled believed a third of the works featured in the auction were trained on copyrightable works because of the generative AI models used. The training of generative AI models remains ambiguous given our existing legal framework has not adjusted to technology disruptors like ChatGPT. Some of the first copyright infringement generative AI cases, Richard Kadrey et al. v. Meta Platforms, remain in discovery, not yet adjudicated, leaving the art world in purgatory on how to guide the use of generative AI in creative spaces. Courts in the United States are examining how generative AI is trained, looking into the details of AI models being able to “memorize,” “synthesize,” and generate near verbatim results of their copyrightable works. Predictions of divergent adjudications in U.S. courts creates concern of consistent jurisprudence when providing guidance on ethical use in the art world. The utilization of generative AI to create art will be significantly impacted as judicial decisions from pivotal generative AI cases are determined.
Conversations about AI are fragmented, quickly dissolving into pro-AI and anti-AI spaces, creating difficulties in exploring the ethical uses of AI. To ensure longevity, building a shared language to describe and understand where to focus on uses of AI in creativity spaces is necessary. The courts may be able to help with this. As generative AI cases are discussed and adjudicated, technical nuances defining key terms like “training” (versus “creating”) and “output” (versus “imitation”) may finally be defined in its applicability to creative spaces, lending toward a universal dialogue in the art world when solving for generative AI training concerns.
Perhaps there is a world where both the artist and AI can win. A collaborative world where the artist receives compensation in the form of licensing or royalites from a service provider offering a library of creative works for generative AI models to be trained from to grant benefits to copyright holders for their creativity. This could allow for a harmony between the two: a space where an artist can continue to produce creative work, and AI helps to further their work.
Student Bio: Kelsey Shaughnessy is a third-year law student at Suffolk University Law School with a passion for Intellectual Property Law, particularly as it intersects with science, art, and design. She is a staff member for the Journal of High Technology Law, where she explores the complexities of protecting creative rights in the digital age. She received a Bachelor of Arts in Finance and Economics from the University of Massachusetts in Amherst.
Disclaimer: The views expressed in this blog are the views of the author alone and do not represent the views of JHTL or Suffolk University Law School.
