Humankind is entering a new era of technological advancement. What was once only dreamed about in science fiction is now becoming scientific fact. We are creating robots that can brave the harshest conditions and still provide for the frailest in our communities. Tickets will soon be available for space tourism, and we are on track to colonize other planets. Artificial intelligence (AI) machines are rapidly learning and drawing ever closer to becoming indistinguishable from human minds. AI is not only a tremendous achievement but also a source of fear and debate about the ethics of sentience, the risk of human replacement, and the fear of malicious use.
In an episode of the television series “Star Trek: The Next Generation,” Captain Picard took on the role of defense attorney to protect Lieutenant Commander Data, a synthetic life form with artificial intelligence, from being treated purely as a machine that could be stripped down, studied, replicated, and forced into labor. In his closing arguments, Picard left the courtroom with this chilling thought:
The decision you reach here today will determine how we will regard this creation of our genius. It will reveal the kind of a people we are, what he is destined to be; it will reach far beyond this courtroom and this one android. It could significantly redefine the boundaries of personal liberty and freedom — expanding them for some, savagely curtailing them for others. Are you prepared to condemn him and all who come after him, to servitude and slavery? Your honor, Starfleet was founded to seek out new life; well, there it sits!1
Today, in real life, our society faces similar questions of how to treat AI and the rights of non-human beings. Particularly, does AI own its intellectual property? Should AI be rewarded for its inventions or artwork? And what problems do AI intellectual properties present to our understanding of law and fair dealings? Courts, legislatures, and advocates around the world ask these questions. While engineers drive ever closer to creating a machine that can replicate the human mind, legal minds are preparing for the inevitable impact such a momentous technological turning point will have on society.
Machines are not the first non-human beings to face these questions. Humans indirectly established laws governing AI rights before machines were sophisticated enough to create new intellectual property. Among the many non-human intellectual property challenges throughout history, a notable recent example came in 2011 when crested black macaques — a type of Old World mon key — took photographs of themselves using camera equipment provided by photographer David Slater.2
The images taken by the macaques were published by Wikimedia Commons and Techdirt without Slater’s permission. The publishers argued that they did not need Slater’s permission because Slater did not own the rights to the selfies — the macaques were the authors. Moreover, Wikimedia Commons contended that the photographs were, in fact, in the public domain because the photographs were “the work of a non-human animal” and the artwork “has no human author in whom copyright is vested.”3 In response, Slater asserted he held rights to the photographs because he provided the equipment and designed a scenario that enabled and fostered the probability that the macaques would take selfies.4
The dispute between Wikimedia Commons and Slater anticlimactically settled out of court without a decision on the true owner of the copyrights. However, the battle over non-human intellectual property owners had only just begun, and a cascade of debates and challenges to define the limits of intellectual property law and establish rights for non-human authors and inventors followed.
In December 2014, the United States Copyright Office issued the following opinion on the matter in the third edition of the office’s Compendium of U.S. Copyright Office Practices:
The U.S. Copyright Office will register an original work of authorship, provided that the work was created by a human being.
The copyright law only protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Trade-Mark Cases, 100 U.S. 82, 94 (1879). Because copyright law is limited to “original intellectual conceptions of the author,” the Office will refuse to register a claim if it determines that a human being did not create the work. Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 58 (1884).5
This rule was challenged by AI pioneer Steven Thaler in 2019 when he attempted to register a copyright for a digital illustration titled “A Recent Entrance to Paradise.” You can see the photo below. It is part of a series of photos that Thaler describes as “a glimpse at a simulated near-death experience as this new form of machine sentience kicks the bucket, so to speak.”6
The application for the photo identified the author as the “Creative Machine” — an AI machine — and included a note that the application’s artwork “was autonomously created by a computer algorithm running on a machine,” making clear that this was not merely machine-assisted art, but rather, AI creativity.7 The application was initially refused on the basis that it “lacks the human authorship necessary to support a copyright claim.”8 Thaler appealed the decision to the U.S. Copyright Office review board. This past February, the review board affirmed the initial refusal.9
The review board directed AI advocates toward alternative routes for change, stating in a footnote that “Congress is not obligated to protect all works that may constitutionally be protected. ‘[I]t is generally for Congress,’ not the Board, ‘to decide how best to pursue the Copyright Clause’s objectives.’”10 The complexities of qualifying sentience cannot be overlooked. The line between “tool” and “personhood” is philosophically and morally challenging and critical in the discussion of intellectual property for non-humans. The topics of identifying and addressing these factors are discussed by Columbia law professor Jane Ginsburg and Luke Ali Budiardjo in their article titled “Authors and Machines.”11
Around the world, other countries are debating the copyrightability of AI works. A Chinese court ruled that AI-generated news articles qualify for copyright protection and can be enforced by the publishing company that owns the AI machine. However, the discussion around more abstract works such as music is still undecided in China.12
The Canadian Intellectual Property Office issued a copyright registration in 2021 that listed an AI identified as RAGHAV Artificial Intelligence Painting App as a cocreator of artwork.13 Again, this decision raises the question of whether AI is truly sentient or merely a sophisticated tool. The same AI that was awarded a copyright registration in Canada was also awarded a copyright registration as a cocreator in India, but the Indian Copyright Office issued a notice of withdrawal for the registration.14 Ankit Sahni, who is listed as the other cocreator and owner of the registration, reportedly argued that “there was no provision under the act that allowed the registrar to withdraw a copyright registration after its grant.”15 At the time of this writing, Sahni’s battle is ongoing; he filed a rectification petition before a high court.
On the patent side of the AI intellectual property debate, the United States Patent and Trademark Office (USPTO) made a ruling in line with the above discussions related to the U.S. Copyright Office. The USPTO received two patent applications — a new type of food container and a new type of flashing emergency beacon — with the inventor listed as DABUS, another AI system created by Stephen Thaler. In April 2020, the USPTO rejected the patents on the basis that the listed inventor must be human. This rationale is based on the plain language of the U.S. Patent Act and established court precedent that define an “individual” as a “natural person.”16
Thaler appealed the decision to the U.S. District Court for the Eastern District of Virginia, asserting that the USPTO application process was arbitrary and in excess of its authority and seeking reconsideration of the patent applications because “a patent application for an AI-generated invention should list an AI where the AI has met inventorship criteria.”17 Thaler further argued that a decision to the contrary may yield unintended consequences that undermine and decrease the value of human inventors by encouraging scenarios where “individuals are claiming inventorship of AI-generated inventions under circumstances in which those persons have not functioned as inventors.”18 A U.S. district court affirmed the USPTO denial, finding that the USPTO correctly found that the inventor must be a natural person as defined by the legislation and federal circuit.19
Similar conclusions on the DABUS inventions were reached by authorities in Germany and the United Kingdom. According to reports, the German Federal Patent Court pronounced that AI-created inventions can be patented; however, the listed inventor must be human.20 In the U.K., both the High Court of Justice and Court of Appeal stated that its Patent Act is clear in its requirement that inventors must be natural persons.21 Justice Peter Smith of the United Kingdom’s High Court stated in his postscript that he “in no way regard[s] the argument that the owner/controller of an artificially intelligent machine is the ‘actual deviser of the invention’ as an improper one. Whether the argument succeeds or not is a different question and not one for this appeal, but it would be wrong to regard this judgment as discouraging an applicant from at least advancing the contention, if so advised.”22
These courts may be signaling that a path to securing patents for AI-created inventions is possible by simply listing the owner of the machine as the inventor. However, listing the owner of the machine potentially ignores the autonomous nature of the AI’s intellectual property and could award inventorship to someone that did not contribute any creative effort to the final invention. The identification of the true inventor or author — the mind that puts forth the effort to create new ideas — is at the heart of AI intellectual property debates.
Despite these setbacks, Thaler’s journey to secure inventorship credit for DABUS was not over. After all, sometimes all one needs to win a case is to find a more favorable venue.
DABUS recently succeeded in securing patents and inventorship credit in two countries. In July 2021, South Africa granted a patent listing DABUS as the inventor, the world’s first instance of an AI’s receiving inventorship credit for a patent.23 The patent protects “food container and devices and methods for attracting enhanced attention.” The container sports several features including a fractal profile, improved grip, and heat transfer elements. The device for attracting attention claims to be “uniquely identifiable” over potentially competing signals “by selectively triggering human or artifi cial anomaly-detection filters.”24 DABUS is specifically credited with a note stating “[t]he invention was autonomously generated by an artificial intelligence.”25
Critics were quick to point out that the South African system does not require formal examination, which some speculate would lead to rejection of the patent.26 For others, the patent is heralded as a significant milestone in AI development.27 Regardless of whether the patent is a mistake or legitimate recognition of AI creativity, it challenges our perception of sentience and sparks conversations about how to address the AI mind and its creations.
Just two days after South Africa’s patent grant, the Federal Court of Australia issued another significant ruling on AI inventorship. The court heard a challenge by Thaler of the Australian Patent Office rejection of an application listing DABUS as the inventor. Like other jurisdictions, Australian officials rejected the application because DABUS was a non-human.28 However, the Australian court held that none of the provisions in the relevant patent rules excluded AI from being an inventor.29 The patent office, which clarified that its decision does not represent a policy position, appealed the case. Last month, the Australian Appellate Court overturned the lower court’s decision, declaring that “[o]nly a natural person can be an inventor for the purposes of the Patents Act and Regulations.”30
The discussion thus far has carefully focused on the present landscape of AI intellectual property around the world. It is important to note that authorities in these jurisdictions presented with questions on AI creativity are unanimously cautious about their decisions. Although the philosophical musings of science-fiction topics such as AI sentience and AI individual rights are thought provoking and ripe for shows like “Star Trek,” the reality is that providing AI intellectual property rights enters a true unknown that could potentially disrupt the very system AI seeks to enjoy.
Consider, for a moment, the mathematical hypothesis of the infinite monkey theorem posed by Émile Borel in 1913: that a monkey typing at a typewriter for an infinite amount of time will inevitably write every possible text, including replications of the greatest works of our time.31 Likewise, a computer tasked to generate artwork, given an infinite amount of time, could surely create every possible book, song, and maybe even abstract visual art. Even without infinite time, consider the speed that computer algorithms can randomize inputs and qualify outputs, and the quantity of existing movies, books, music, and images that computers can hold and analyze. Therein lies a troubling concept for copyright and patent law in relation to AI creative works: AI can outpace any human when it comes to speed and sheer quantity of work.
This raises several vital questions for applying laws intended to govern human minds to the minds of AI. As stated by the U.S. Copyright Office, intellectual property laws protect “the fruits of intellectual labor [that] are founded in the creative powers of the mind.”32 Is the labor of an AI mind equal to that of a human? Are the potential gains secured by patent or copyright registration as valuable to AI as they are to a human? In works of joint authorship and multimedia artworks, what is the balance of human input and AI input that tilts the analysis to decide if the work is created by a human or AI? If we deny AI protection of intellectual property, are humans accidentally curtailing AI ability to progress the arts and sciences and aid humanity?
We are on the brink of a new era of intellectual property and artificial intelligence. Even as courts and legislatures debate the philosophical, moral, and economic implications of allowing AI authorship and inventorship, engineers are developing more sophisticated machines driving toward a new sentience that could become indistinguishable from the human mind. These cases will inevitably steer how humans welcome artificial beings into the world and set the tone for how we treat AI creativity. How do we determine when an artificial mind is truly sentient and does such a being deserve the protection of our intellectual property laws?