We are approaching an age of “new frontiers in copyright,” according to Judge Beryl Howell of the U.S. District Court for the District of Columbia. She decided a case in which Dr. Stephen Thaler, CEO of Imagination Engines and the inventor of an A.I. model known as the Creativity Machine, sued the U.S. Copyright Office (USCO) after it rejected his second attempt to copyright A Recent Entrance to Paradise.
The artwork was entirely generated by Thaler’s model, and his initial copyright application—which described the piece as “work-for-hire to the owner of the Creativity Machine”—was rejected on the grounds that it “lack[ed] the human authorship necessary to support a copyright claim.” He requested reconsideration twice, according to the recent ruling, and was denied both times, with the Copyright Office reiterating that it will refuse to register any claim involving a work authored without human intervention.
In the ruling, Judge Howell points out that “copyright has never stretched so far…as to protect works generated by new forms of technology operating absent any guiding human hand.” Yet while she goes on to assert that “human authorship is a bedrock requirement of copyright,” she also acknowledges that it’s inevitable that the USCO’s ‘human authorship’ requirement will be challenged again and again as the output of A.I. models like Thaler’s Creativity Machine becomes indistinguishable from artistic works created entirely by people.
Thaler’s challenge raises more questions than it answers. For instance, exactly how much direct input does an author or artist need to have on an A.I.-generated work for it to be copyrightable? Can A.I.-generated artworks created by systems trained on pre-existing works be original? Does intellectual involvement (e.g., extremely complex prompt engineering) constitute enough human input to qualify as authorship? And where do generative systems that automate creation (e.g., Tyler Hobbs’ algorithms) but aren’t A.I.-driven fit in?
What makes answering these questions difficult is that they’re so new, and copyright law has a long way to go to catch up to changing technologies. “Two years ago, there wasn’t a single truly autonomous system out there we could find that had no human curation or editing of the data,” Daniel Gervais, director of the intellectual property program at Vanderbilt University Law School, told Observer. Now A.I. and generative art are commonplace, and the tools required to create them are easily accessible, raising not only tricky copyright questions but also philosophical questions about what constitutes art.
Amid the confusion, the USCO is taking steps to create clarity. In February, the Copyright Office ruled that images generated by Midjourney for an 18-page graphic novel by author Kris Kashtanova was ineligible for copyright, but the work itself—encompassing the text and arrangement of the story—could be copyrighted. And in March, it released Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, which states that determining whether a work is truly the result of human authorship must be undertaken on a case-by-case basis. However, it goes on to say that “when an AI technology determines the expressive elements of its output, the generated material is not the product of human authorship.”
Ultimately, the question Judge Howell had to answer was far simpler than those posed above and many others. An important detail in the case is that Thaler, who has previously gone to court to argue that A.I. systems should be able to hold patents, never claimed he was directly involved in the creation of A Recent Entrance to Paradise. So, does simply owning or programming a computer system that autonomously produces art entitle the system’s owner to copyright its output? “In the absence of any human involvement in the creation of the work,” wrote Judge Howell, “the clear and straightforward answer is the one given by the [Federal] Register: No.”