The intersection of artificial intelligence and copyright law

There’s a gap in between

There’s a gap where we meet

Where I end and you begin

And I’m sorry for us

The dinosaurs roam the earth

The sky turns green

Where I end and you begin

Radiohead, “Where I End and You Begin”

Perhaps Radiohead was not thinking about artificial intelligence and copyright when they wrote “Where I End and You Begin” back in 2003, but their words foreshadowed a generational fault line in the law.

AI image generators are computer programs that create pictures based on a text prompt provided by a user. For example, if a person enters the prompt, “imagine a lizard on a rock in the desert at sunrise,” the image generator will create a picture of that scene in a matter of seconds. Image generators don’t simply copy and paste bits of prior works into a Frankenstein-style new image. Instead, most image generators go through a process of comparing random images to real images until they come up with rules about how things look. Based on those rules, they can create new images guided by text prompts provided by users.

Stephen Thaler developed just such a system, which he called the “Creativity Machine.” Thaler used that system to generate an image that he titled, “A Recent Entrance to Paradise.” Thaler then tried to register the image with the United States Copyright Office. In his application, Thaler explained that the work had been “autonomously created by a computer algorithm running on a machine.” The Copyright Office denied Thaler’s application on the ground that the work “lack[ed] the human authorship necessary to support a copyright claim.” The Copyright Office took the position that copyright protection only applies to works created by a human being. Because Thaler admitted that this work had been generated by a computer and lacked any human authorship, it was not protectable under copyright law.

Thaler sued the Copyright Office in the United States District Court in Washington, D.C., seeking to have the Court overturn the determination and grant him a registration. The Court refused to do so. On cross-motions for summary judgment, the court ruled that “In the absence of any human involvement in the creation of the work, the clear and straightforward answer is the one given by the Register: No.” The Court relied on prior cases holding that images created without human involvement could not be protected by copyright. Most recently, in 2018, the Ninth Circuit refused to grant protection for a photograph taken by a crested macaque because animals lacked standing under the Copyright Act. Naruto v. Slater, 888 F.3d 418, 420 (9th Cir. 2018).

The Court went beyond historical precedent to the underlying rationale for denying copyright protection to works created by AI. The Constitution itself provides the authority for Congress to enact copyright laws 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.” U.S. Const. art. 1, cl. 8. The purpose of intellectual property protection is to incentivize individuals to create and invent. But as the Court astutely pointed out, “Non-human actors need no incentivization with the promise of exclusive rights under United States law, and copyright was therefore not designed to reach them.”

This particular case is very straightforward because Thaler admitted that the work that he applied to register was entirely the creation of the machine. Of course, many AI creations are not that simple. Authors may use AI tools to create drafts, which the authors then edit or re-write. Artists may edit works generated by AI, leaving it less clear how much of the work is a human creation. In a recent case, Kris Kashtanova submitted a graphic novel to the Copyright Office for registration. Kashtanova wrote the story and arranged the artwork but admitted that the images were all AI generated. The Copyright Office granted registration, but limited it to the text and arrangement, excluding the images.

As AI tools become more prevalent, the line between human authorship and machine creation is becoming more blurred. Science fiction author Tim Boucher made headlines recently for writing 97 books in 9 months by using AI tools. Boucher says he used ChatGPT and Anthropic’s Claude for brainstorming and text generation. Stephen Marche says he composed a murder mystery book, Death of an Author, using three AI tools. Marche himself initially wrote an outline with prompts and notes. He then ran it through ChatGPT to create a story based on his outline and notes. Marche used an AI called Sudowrite to adjust the style to make the prose closer to his writing. Finally, he used an AI system called Cohere to tweak the writing, adding similes and other flourishes. Though the underlying plot was clearly Marche’s, much of the writing was not.

It is that gap where man and machine meet, where the human ends and the machine begins, that will be the most vexing question for the law. It is easy to say that human creations are protectable and machine-made images are not. The two will increasingly work symbiotically, creating a gray area between what is protectable and not. The Thaler ruling is merely a first signpost on what will be a very long road into a murky area for copyright law and for authors. It may ultimately take an act of Congress to sort out, but we can take hope in the guiding star identified by the Court in the heart of this decision. Ultimately, the goal is to spur creation and innovation, whether entirely human or somewhere in between.