« Features

Artificial Intelligence and Copyright

When AI is involved in the authorship of a work, how does it affect the copyright?

By Octavio Robles, AIA, Esq.

Under U.S. law, the ownership of copyright to a work accrues to the author of the work, regardless of the medium used. That is always the case with the exception of works-for-hire, where ownership is effectively assigned to an employer. This policy of authorship protection is rooted in the idea that in order to incentivize people to be creative, a certain degree of protection against copycats is necessary in order to allow the author time to benefit exclusively from his ingenuity. Our founding fathers made sure that creativity was promoted and protected when the U.S. Constitution was enacted. Article I of the Constitution gave congress the power to “…promote the progress of science and useful arts by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries.” Since then, copyright and patent law has evolved, under the direction of congress, in order to respond to the development of technologies and acceptable social standards. Sometimes congressional action lags behind and a vacuum arises when the law does not keep up with technology. That may be the case at the present time when it comes to Artificial Intelligence (AI) and intellectual property protection in the United States. This article focuses on how the recent advances in AI affects copyright.


From the beginning of technological advances during the industrial revolution in the 19th century, machines that helped humans to be creative and develop a resulting product, were considered tools of the human authors under then established copyright law. One of the earliest cases regarding copyright protection for a product that used a technological “tool” was the 1884 case of Burrow-Giles Lithographic Co. vs. Sarony. This was the landmark U.S. Supreme Court case that first extended copyright protection to photography. The Court held that the camera was merely a tool that captured the image of writer Oscar Wilde by photographer Napoleon Sarony and resulted in a copyrightable work of art. The case is still relevant today as valid legal precedent. Fast forward to the present where computer technology is commonly used as a tool of any given methodology which is programmed by humans to obtain a pre-determined result.

Norwegian Pia MYrvoLD uses the input from non-passive art spectators through computer programs to produce randomly created images that respond to the public’s participation and interaction. Installation view of her exhibition “ART AVATAR,” Centre Pompidou, Paris, 2014. Photo: Anya Buklowska.

Norwegian Pia MYrvoLD uses the input from non-passive art spectators through computer programs to produce randomly created images that respond to the public’s participation and interaction. Installation view of her exhibition “ART AVATAR,” Centre Pompidou, Paris, 2014. Photo: Anya Buklowska.


As technology has developed, computers have become an important part of the mediums available to someone creating a work, whether visual or literary. As computers and other electronic devices have become more sophisticated, they have become able to create works. The works created by such devices have been mostly enabled through programs and directives of a human. In those cases, the copyright accrues to the person behind the program or directive, unless it was a work for hire. In such instances, the copyright would accrue to the person or entity “employer” that would own the resulting product of the device. Where AI is used as a methodology which is again, programmed by a human but instead, is designed to respond to the input of other humans and the resulting product is a reflection of those multiple inputs, then, in such cases, the copyright accrues to all humans that provided an input through either the initial programming or via the subsequent computer “tool” stimulations. This is referred to as interactive creativity.


The Norwegian artist Pia MYrvoLD is a pioneer of Interactive Art. Among many of her works, in an exhibition called “ART AVATAR,” at the Pompidou Center in Paris in 2014, the artist used the input from non-passive art spectators through computer programs to produce randomly created images that responded to the public’s participation and interaction. The images were ultimately displayed within the actual exhibit. In interactive situations like this in the U.S., the input from the spectators grants each a resulting copyright in a “derivative” work as exhibited by the computer program, which is a “tool”. In order to participate and not proliferate rights under copyright law, each spectator should first agree, in writing, to waive any right they might obtain to a resulting derivative work due to their input. This would have the same effect as having all participating spectators doing work-for-hire and the entirety of the copyright accruing to the artist who created the program.


The whole issue of where does copyright accrue becomes much more complicated when a computer is programmed or instructed to “learn” and “think” by itself. In such cases, the computer has what is referred to as Artificial Intelligence or AI for short. Where the works created by the AI machine is totally independent of and distanced from human control and direct input, theoretically, the copyright would accrue to the AI machine, the author. U.S. law at present, does not provide for copyright protection to accrue to non-humans. Therefore, any work that is copyrightable that does not accrue to a human, automatically falls into the public domain. This is a dilemma that can have a significant negative effect on creativity and the use of AI. In the 2011 case of Naruto vs. Slater, a British wildlife photographer by the name of David Slater was in Indonesia to photograph the Macaques monkeys of the area. He intentionally left a camera on a tripod with a remote shutter accessible to the monkeys. One of the monkeys took advantage and proceeded to take a number of photographs of itself. The “monkey selfies”, as they were referred to became a sensation around the world. When Slater tried to license them under the presumption that he held the copyright, he was sued in a U.S. Court by the group People for the Ethical Treatment of Animals (PETA) which named the monkey in question, “Naruto” and claimed, among other things, that Slater did not own the copyright, Naruto did. The court held that neither Slater or Naruto owned the copyright. In the case of Slater, because he did not take the photos. In the case of Naruto, because it is a monkey and therefore, non-human, thus lacking standing in Court to claim the copyright even though it is technically, the author. The case was dismissed and the copyright to the “Monkey Selfies” fell to the public domain.


There are many disadvantages to having AI created art fall into the public domain. The programmers and investors that would otherwise benefit from the copyright are discouraged from moving ahead with cutting edge AI generated works. This situation is exactly what the founding fathers did not intend and what congress has demonstrated over the years that it does not want. It is important to understand the distinction between authorship and ownership of copyright. A legal entity can and often does own copyright. Human authors who initially hold the copyright are able to assign their rights to the copyright to any legal entity. However, in order to transfer same, a human has to acquire it first. There are two possible approaches to solving the AI copyright dilemma. One requires congressional legislation, which makes it more difficult to achieve, that changes the Copyright Act to allow non-human entities, like corporations and LLCs to be considered “authors” and initially acquire copyright. This alternative has legal ramifications that would need to be evaluated and are beyond the scope of this article. The other alternative is to allow a more flexible interpretation of the terms “employee” and “employer” as it may apply to the DI machines that would do “work-for-hire” whereby the ownership of the copyright never accrues to the DI machine and therefore does not fall into the public domain. Rather, the ownership accrues directly to the “employer”, which could be any legal entity or a human. This may be achievable as an interpretation of the existing Copyright Act, which is much more easily attained.

Octavio Robles, AIA, Esq. is a legal contributor to ARTDISTRICTS and a member of the Florida and Federal Bars (Southern District of Florida). He is a Florida Supreme Court Certified Mediator and Approved Arbitrator; a member of the Construction Committee of the Real Property, Probate and Trust Law Section, the Art and Entertainment, and the Alternative Dispute Resolution Law sections of the Florida Bar; and a member of the American Institute of Architects, Art Deco Society of Miami and Copyright Society of the USA. He holds licenses as a registered architect, state-certified general contractor and real-estate broker in Florida and is a LEED-accredited professional. He received his Juris Doctor degree from the University of Miami School of Law in 1990. He holds masters’ degrees in architecture and construction management and a bachelor’s degree in design, all from the University of Florida. His practice is limited to art, design, architecture, construction and real-estate law.