In recent years, the news has been flooded by innovations in artificial intelligence and its ability to create a diverse range of creative content. AI has produced musicpoetrypaintings, and even sci-fi films. Some of the poetry is so “real” that it has even fooled humans into believing that it was generated by a human writer. As the list of creative works continues to grow, so does the elephant in the room: who owns the copyright in AI-produced works? Similar issues have arisen in the patent context. Answering this question becomes especially treacherous when it comes to creative works based on machine learning, a method of AI in which the AI learns from and creates works based on a pool of data.

The Next Rembrandt project (the “Project”) uses machine learning based AI to mimic the works of the Dutch artist.  The Project’s goal is to answer the question, “can the great master be brought back?” The Project examined the entire collection of Rembrandt’s work, studying each pixel of his work in high resolution 3D scans and digital files. In order to maximize the pool of data from which the AI would draw, the Project narrowed its exploration to portraits, Rembrandt’s most popular subject. The Project further specified that the AI would create a portrait of a “Caucasian male with facial hair, between the ages of thirty and forty, wearing black clothes with a white collar and a hat, facing to the right,” analyzing Rembrandt’s brush strokes to determine the exact angle and number of layers of ink that should be applied. The result is “original” pieces created by AI in the “style” of the Dutch master. However, this conception of originality seems incompatible with the notion of copying another’s style. Gut instinct might tell us that something based entirely on recreating another’s work feels entirely unoriginal. Should Rembrandt be the owner of the work, the person(s) who created the AI, or the AI itself?

 

The Copyright System and The Romantic Author

The current US Copyright Office policy is to reject copyright claims for works produced by non-humans, releasing them into the public domain. This policy stands despite the fact that there is no explicit requirement for human authorship in the US Copyright Act. In her article, Coding Creativity: Copyright and the Artificially Intelligent Author, Professor Annemarie Bridy states that courts view authorship as a purely human phenomenon. She expresses the concern that this may be the result of society’s prevailing conception of romantic authorship, leading to the formulation “the threshold requirements for authorship in terms of mind and intellect.”

Arguments made by both parties in a current case in California involving a rare macaque monkey named Naruto, reflect this romantic sentiment. In 2011, Naruto picked up British photographer David Slater’s camera and took a selfie. Slater claims to own copyright in the images though he  neither held the camera nor did he click the shutter. He asserts that he was responsible for setting up the shot—he chose where to leave the camera and arranged it in a manner that Naruto could take the photograph. People for the Ethical Treatment of Animals (PETA) brought suit on behalf of Naruto, claiming he is in fact the copyright owner in the images. According to PETA’s complaint, Naruto examined and manipulated Slater’s unattended camera by “independent, autonomous motion” and by “purposely pushing” the shutter release, “understanding the cause-and-effect relationship between pressing the shutter release, the noise of the shutter, and the change to his reflection in the camera lens.” Both arguments demonstrate the idea that the outcome of competing authorship claims may be determined by whoever can best show that they are the mind and intellect behind the resulting work.

The District Court granted Slater’s motion to dismiss, remaining silent on whether Naruto had standing under Article III. The District Court simply concluded the Copyright Act does not confer standing on animals, reasoning that the Act did not “plainly” extend the concept of authorship or statutory standing to animals. Though Naruto is not an artificial being, this case may provide some insight into how attempts to acquire intellectual property in AI will be treated in the future. The court seems to be steadfast in the idea that copyright authorship is reserved for legal persons.

If we accept that copyright ownership only applies to human beings, then who might own copyright in The Rembrandt Project? Perhaps the creators of the AI should own the copyright. After all, it is their intellectual labor that made the works created by AI possible. In that sense, the creations produced by AI could be thought of as derivative works.  This approach could run into some problems because not all derivations are subject to copyright protection. For example, copyright will ideally not protect stock characters from a fictional story regardless of its originality. Alternatively, it may be most desirable to assign copyright to Rembrandt because every brush stroke and creative choice made by the AI is informed by Rembrandt’s past stylistic choices. But, this approach may run counter to the aim of copyright because it provides ownership rights to a work that the owner took no part in.

 

“To promote the progress of science and useful arts…”

Questions regarding copyright in AI works may be best answered by taking a step back to think about how its protection fits within the broader goals of copyright. Article I of the US Constitution expresses states that “Congress shall have 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.” This clause supports the notion that innovation is the driving force behind human progress. The exclusive right provided prevents others from using IP and provides the owner with the ability to generate a benefit. In that vein, one reason that we may be concerned about AI copyright ownership is that AI does not have any of those needs nor can it take advantage of IP benefits. However, the threat of AI copyright narrowing the playing field for innovation may create a more competitive market for expression, thus advancing human progress. AI in The Rembrandt Project, for example, is able to analyze works of a great master in a way the average art student may not be able to such that it may drive art students to study and produce works in a whole new way.

Article I also brings up larger questions about the goal of the arts and why humans place so much value on it. The question of AI authorship may, in fact, be more of a philosophical inquiry about what it means to be human and what boundaries we put on that classification. What would it mean for our sense of the human condition if we allow non-human entities to share the same intellectual rights? And what does that mean for the laws that govern human action if they also regulate non-human action?