The A.I. Wars are almost here. No, I’m not talking about Terminator or even a crackdown on Twitter bots. Instead, we’ll soon be witnessing a series of extraordinary test cases designed to force the American legal system to reconsider the concept of authorship as artificial intelligence begins to write short stories or pop songs. It may sound like a Zuckerbergian fever dream, but A.I. could soon be creating blockbuster movies and life-saving pharmaceuticals, too—multi-billion dollar products with no human creator.
The legal battle has already begun. Sometime in the next couple of weeks, I’ve learned, a lawsuit will be filed that challenges the U.S. Copyright Office’s recent decision to deny an “author” identified as “Creativity Machine.” Then, a few weeks later, a federal appeals court will hear oral arguments in Thaler v. Hirshfeld, an under-the-radar but potentially blockbuster case concerning whether A.I. can be listed as the “inventor” in a patent application. Meanwhile, authorities in the European Union and 15 other countries are being asked to make similar determinations to properly credit the achievements of A.I.
The instigator for all this action is a Yale-trained lawyer named Ryan Abbott, a globetrotting, modern-day renaissance man who has recruited Dr. Stephen Thaler, a pioneer in the field of A.I. who developed the Creativity Machine and other modern engineering marvels, as his model client. Abbott, a partner at the Los Angeles-based firm Brown Neri Smith & Khan, also has a medical degree from the University of California, San Diego, and is a scholar. Besides practicing law, both in the United States and England, he teaches at the University of Surrey School of Law and the David Geffen School of Medicine at UCLA.
Most importantly, Abbott is obsessed with our technological future. In writings including The Reasonable Robot, he outlines how we should discriminate between A.I. and human behavior under the law. For example, if businesses get taxed on the wages of its human but not robot workers, he asks, do we incentivize automation? And if we hold the suppliers of autonomous driving software to a punishing tort standard (i.e. strict liability rather than negligence), will there come a time when we’re actually discouraging the adoption of technology that would prevent accidents on the road?
Currently, the topic that Abbott is pushing courts to clarify is the relationship between A.I. and intellectual property. In a nutshell: Must government agencies accept and properly record the ingenuity that’s coming from our increasingly sophisticated machines? Abbott argues yes and believes it’s important that this happens sooner rather than later. “People always say that technology lags behind the law,” he told me. “This is an opportunity for the law to play a role in the development of technology.”
Accordingly, Abbott will soon be filing a lawsuit in federal court that appeals the Copyright Office’s rejection of a registration for “A Recent Entrance to Paradise,” a two-dimensional artwork created by Dr. Thaler’s Creativity Machine. In a Feb. 14 determination (read here in full), the Copyright Review Board told Abbott that the work was ineligible for registration because “human authorship is a prerequisite to copyright protection in the United States.” What’s more, the Board threw cold water on the controversial proposition that the work-for-hire doctrine—whereby an employee assigns authorship rights to an employer like a corporation, which, Abbott points out, is a non-human entity—opens a similar path for the owners of AI. “The ‘Creativity Machine’ cannot enter into binding legal contracts,” stated the Copyright Board in an assessment that will now be reviewed. “[T]he work-for-hire doctrine only speaks to the identity of a work’s owner, not whether a work is protected by copyright.”
This new case focusing on copyright protection for A.I.-generated work could become meaningful for the creative industry as studios and filmmakers explore A.I.’s potential. In recent years, for example, Warner Bros. has used A.I. to guide its decision-making about what projects to pursue. In Japan, a new film about a boy’s dislike of tomatoes, based on a script by A.I., is now hitting the festival circuit. There’s now an A.I. tool out there that, sensing the tone of any video, recomposes music for a score. Sony, in fact, has tried to use A.I. to make new music that sounds like The Beatles, and Spotify is experimenting too. And as anyone who has seen the deepfake “Tom Cruise” knows, A.I. can do a pretty good job of replicating actors (something that’s of increasing concern to actor unions). Put it all together, and we’ll likely be seeing A.I. acting soon as the auteur on a major motion picture. And not just for movies either. A.I. is increasingly involved in video game development, too.
The implications are immense. For example, at least one scholar has pointed out that Spotify could use A.I. creation to develop tunes it would then own, which could diminish the pot of royalty payments for human musicians. Abbott is mindful of such talk. In one law review article, he writes, “If A.I. does not have rights and a human author is no longer in the picture, is there still a case for copyright if we are no longer concerned with protecting authors in this context? Do we still want to apply copyright the same way? Do we need human centric laws? Do the laws that currently apply to people also change once A.I. is in the picture?”
While this will be the first lawsuit that directly confronts a claim of A.I. authorship, courts have been exploring the derivation of creativity and deciding which works get protected for quite some time. Back in 1884, for instance, the Supreme Court tackled whether photographs exhibited sufficient originality or were mere mechanical reproductions of their subjects. A few decades later, a British justice had to wrestle with a copyright case that involved séances and spirit-guided “automatic writing.” Authorship by non-human spiritual beings would come up in U.S. courts, too. Not to mention supposed authorship by Jesus himself.
More recently, and rather famously, a court had to tackle whether a “monkey selfie” was eligible for copyright protection. In fact, when that case (Naruto v. Slater) was being decided a few years ago, the PETA lawyer who was ostensibly representing the monkey told the court that the issue of non-human authorship presented in the case would impact the future of artificial intelligence. In the end, a judge concluded that monkeys don’t have standing to go to court to enforce legal rights, leaving no definitive answer to the question of whether animals can be authors of copyrighted work. Hey, guess what? Abbott’s forthcoming case may settle that issue as well.
Scholars, sometimes at the behest of Congress, have been writing about and debating whether machines can truly be deemed “authors” and “inventors” for decades. For the imaginative academics who have engaged on this topic, the issue harkens to the original purpose for having intellectual property—as the U.S. Constitution puts it, “To promote the progress of science and useful arts.” How should society properly align incentives to encourage innovation and creativity? Some scholars have argued this is a “bad penny of a question” as there tends to be a programmer behind the algorithm. Technology is a tool. Maybe there’s no need to worry about authorship from computers.
From Abbott’s vantage point, there’s a practical purpose to properly crediting the source of creativity—and it has nothing to do with giving a boastful A.I. machine an ownership claim over the fruits of its labor. In fact, Abbott doesn’t want that. He wants the A.I.’s owner to get those benefits. But he worries that won’t be the outcome if there’s no integrity in the patent and copyright systems. As his court papers in his patentable A.I. case put it, “Allowing a person to be listed as an inventor for an AI.-.Generated Invention would not be unfair to an A.I., which has no interest in being acknowledged, but allowing people to take credit for work they have not done would devalue human inventorship.”
Abbott gives an example. He asks me to imagine a pharmaceutical company directing an A.I. to find a cure for Covid. Let’s say the A.I. finds a way to create antibodies. The company then lists its top scientist as the inventor of this novel approach while taking ownership of the patent for itself. “The Patent Office doesn’t challenge when someone says they are an inventor,” he notes. “But a patent can be invalidated if you don’t follow the rules, and there’s certainly a scenario where the scientist is deposed in some patent infringement suit down the line and has to admit having an A.I. solve the critical problem.”
The notion that an important patent could be invalidated some day thanks to the role of A.I. in the development of the invention leads to a key question: Would that pharma company do everything it could to cure a disease if certain paths resulted in no patent protection?
With such uncertainty in the background, Abbott is doing what he can do now to list one of Dr. Thaler’s machines, DABUS (“Device for the Autonomous Bootstrapping of Unified Science”), as the inventor of a beverage container with a design that’s based on fractal geometry. The Patent Office has rejected the attempt. So, in June, Abbott will be appearing before the Federal Circuit and disputing the conclusion that an inventor must be a “natural person.” His primary argument is there’s nothing in the U.S. Constitution nor patent statutes that narrowly excludes artificial intelligence.
Au contraire, responds the nation’s patent authorities, pointing out that the Patent Act uses the word “individual” plus “himself” and “herself.” Sharing gender pronouns? Here, it may be legally significant. As the government argues, “By using personal pronouns… Congress only strengthened the conclusion that it was referring to a ‘human being’ in referencing an ‘individual.’”
The appellate hearing will take place on June 6.