By: The Michelson Institute for Intellectual Property
Executive Editor: David Orozco, J.D., Bank of America Professor at Florida State University & Editor-in-Chief at American Business Law Journal
Artificial intelligence (AI) art is a form of digital art that is generated or manipulated by AI algorithms. These algorithms can range from simple rule-based systems to more complex machine-learning models that generate unique and sometimes unpredictable outputs.
AI art often involves training machine-learning models on large data sets of existing artworks and then using those models to generate new works. The algorithms can be programmed to consider various elements such as color, shape, texture, and composition to generate the output.
How Does AI Art Work?
There are several methods of creating AI art, including:
- Style Transfer: this involves taking a pre-existing artwork and using machine-learning algorithms to apply the style of that artwork to a new, unrelated image.
- Generative Adversarial Networks (GANs): these are machine-learning models that consist of two neural networks – a generator and a discriminator. The generator creates new images, while the discriminator evaluates the images and determines whether they are similar to real images or not. The two networks work together to create new, unique images.
- Evolutionary algorithms: these algorithms use a process of selection and reproduction to generate new images over time. The algorithm starts with a set of randomly generated images and then selects the most visually appealing images to use as the basis for the next generation.
AI art is still a relatively new field, and there is much debate about what constitutes “true” AI artwork. However, the potential for AI algorithms to create new forms of art and explore creative avenues that were previously unavailable to human artists has led to a growing interest in the field.
Concerns and Criticism of AI Art
Critics have raised a variety of concerns about the use of AI in the creation of art, such as:
- Lack of originality: some argue that AI art is not truly original, as it is generated by algorithms that have been trained on existing artworks (without the original artist’s permission).
- The role of the artist: critics argue that if a machine is generating the art, there is no real artist behind the work. This raises questions about the value of the artwork and the role of the artist in the creative process.
- Emotional detachment: some argue that AI art is emotionally detached, as it is created by algorithms that do not have feelings or personal experiences.
- Lack of human touch: others argue that AI art lacks the human touch and personal expression that are key elements of traditional art.
Role of Fair Use
The role of fair use and creative commons in AI art is a complex and evolving issue, as the existing legal framework was not designed specifically for the creation of art using artificial intelligence. However, the holding in Graham v. Prince may provide some insight into how fair use can be applied. Fair use is a legal doctrine that allows limited use of copyrighted material without permission from the owner if the use is for a transformative purpose, such as criticism, comment, news reporting, teaching, scholarship, or research.
In Graham v. Prince, the court considered whether Prince’s use of Graham’s photographs in his own artwork was fair use. Prince argued that his use of the photographs was transformative because he added new elements, such as paint and collage, to the original images. The Court, however, disagreed, holding that Prince had not materially altered the composition, presentation, scale, color palette, and media originally used by Graham.
This case highlights the importance of transformation in determining whether a use of copyrighted material is fair. In order for a use to be considered transformative, it must add something new to the original work, change the original work’s meaning, or create a new aesthetic.
Is AI Art Copyrightable?
In most countries, copyright law protects original works of authorship that are fixed in a tangible form of expression. With respect to AI art, the question of copyrightability may depend on several factors, including the level of originality of the AI-generated work and the role of human creators in the creation process.
Section 101 of the U.S. Copyright Act defines a copyrightable work as “an original work of authorship fixed in any tangible medium of expression.” However, the definition of an “author” is not explicitly provided, and recent case law suggests that an author cannot be a computer. In Naruto v. Slater, the Court reasoned that other sections of the Copyright Act imply that the author must be a human being.
Furthermore, the U.S. Copyright Office will not grant a registration for a work unless the author is a human being. The Copyright Office’s position is that works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author are not copyrightable.
However, it’s worth noting that the copyrightability of AI art is an emerging issue, and it is possible that the position of the U.S. Copyright Office could change in the future as AI technology continues to advance and courts are faced with new and challenging questions about the role of AI in the creation of works of authorship. In fact, as of this month the U.S. Copyright Office has just cancelled, at least partially, the copyright registration granted to the author of “Zarya of the Dawn” comic book due to images produced via Midjourney AI.
Patents & Artificial Intelligence
AI is posing challenges for the patent system, in addition to copyright law. Currently, the EU has explicitly forbidden AI from registering a patent as an inventor, while the USPTO has yet to issue a formal policy on the matter.
Under U.S. patent law, an inventor must be an individual who conceives of an invention and reduces it to practice. However, the Patent Act does not explicitly define “individual.” In Thaler v. Vidal, a Federal Circuit Court held that the definition of an inventor is limited to natural persons, citing “the overwhelming evidence that Congress intended to limit the definition of ‘inventor’ to natural persons.” Thus, it is likely under current law that AI systems, despite their capability to generate new ideas, cannot be considered inventors.
South Africa, on the other hand, has allowed AI to be listed as an inventor on a patent based on the understanding that the AI system in question had significantly contributed to the invention and that the system’s owner should be recognized as the patent holder. This decision could set a precedent for other countries as AI continues to play a more prominent role in innovation and invention.
Can Creatives Fight Back Using IP?
Artists and creatives may use U.S. copyright law to protect their works from unauthorized use or infringement, including works generated by AI. However, the exact extent of protection will depend on the specific circumstances of the case and the application of relevant legal principles, such as fair use and the doctrine of originality.
For instance, if an AI-generated work is deemed to be a “derivative work” based on the original creative work of an artist, the artist may have the right to control the use and distribution of that derivative work. On the other hand, if the AI-generated work is considered a “transformative” use of the original work, it may qualify for protection under the doctrine of fair use, which would allow it to be used without permission from the original artist.
Further Resources
The Michelson 20MM Foundation, The Intangible Advantage, Understanding Intellectual Property in the New Economy
Video, Fair Use Explained
_________
The Michelson Institute for Intellectual Property, an initiative of the Michelson 20MM Foundation, provides access to empowering IP education for budding inventors and entrepreneurs. Michelson 20MM was founded thanks to the generous support of renowned spinal surgeon Dr. Gary K. Michelson and Alya Michelson. To learn more, visit 20mm.org.