Recently the US Copyright Office (USCO) reaffirmed its longstanding position that human authorship is a pre-condition for recognition of copyright, despite being pushed hard by a particular protagonist within the tech community to accept artificial intelligence (AI) in the form of a machine (NB his machine) or software as a creator with rights. Stephen Thaler, PhD (identified by some sources as Steven Thaler) is, according to this website, the CEO of Imagination Engines Incorporated. His LinkedIn profile says that he is “a pioneer in the area of artificial intelligence and the inventor of the Creativity Machine(R) Paradigm, US Patent 5,659,666 and its derivatives.” Apparently, this machine has been used “to design a wealth of commercially available products”. It must be quite a machine because apparently it also produces autonomous art. This art, a two-dimensional image entitled “A Recent Entrance to Paradise”, was the subject of the USCO Review Board’s recent decision, which ended a three-year struggle by Thaler to get his Creativity Machine recognized as the creator of the work.
According to the USCO, in 2018 Thaler sought to register the work, identifying the author as the “Creativity Machine”. Knowing the well-established position of the Copyright Office not to grant copyright to works created by non-human efforts, such as works created through AI, Thaler tried to invoke the process of recordation, whereby owners may transfer their copyrights to others, and have this transfer recorded by the Office. In this case, he claimed that the work “was autonomously created by a computer algorithm running on a machine” and he was seeking to register this computer-generated work as a “work for hire” to the owner of the Creativity Machine, himself. When Thaler’s application was refused, he appealed, arguing not only that the requirement for human authorship was unconstitutional but also that copyright law already allows non-humans (i.e. corporations) to be authors under the “work for hire” doctrine. Although his appeal was denied by the USCO he appealed yet again, this time to the Review Board. In refusing to revisit the Office’s decision to reject the application, the Review Board cited multiple precedents where human authorship was established as a prima facie requirement for copyright registration. As for the “work for hire” argument, it pointed out that a work for hire must be prepared by an “employee” or those who have contracted to produce a work for hire. In both cases a binding contract or agreement is required. A machine cannot enter into a contract, whereas a corporation (a legal entity) is able do so. Moreover, the work for hire doctrine only identifies the owner of the work, not whether it is protected by copyright. And since it was not created by a human, it cannot be subject to copyright. Case dismissed.
One can speculate as to why Thaler was so insistent on imbuing his Creativity Machine, for which he holds US Patent no. 5,659,666, with the attributes of authorship. Could it have something to do with raising the profile and marketability of this invention? You can imagine the coverage. “Art produced by AI Machine first to be awarded Copyright Protection.” Despite Thaler’s persistence, it ain’t gonna happen. Sorry Steven, or is it Stephen, but it was a valiant effort. You should have registered the copyright in Britain, where there is a provision for copyright protection for machine-generated works where no human author of the work is involved. (see below)
Blogger and copyright scholar Stephen Carlisle has speculated that one outcome of the doctrine that stipulates machines cannot be authors will be that humans will claim for themselves works that have been produced by machines. In other words, they will lie to gain the protection of copyright for works created autonomously by AI. It is worth noting that Thaler’s position was exactly the opposite. He went to great lengths to specify that the work for which he was seeking registration was the creation of his Creativity Machine, not him. The Review Board anticipated the argument that denying copyright protection for machine generated works will encourage dishonesty and inaccurate claims of authorship by noting the criminal penalties that exist for those knowingly making false representations, plus the fact that registration can be cancelled if invalidly filed. In Britain a similar concern has been raised since works produced by human authors get a longer term of protection than purely machine-generated works. Like the USCO, the British authority considers that legal provisions regarding fraudulent representations provide sufficient protection against such activities.
The question of whether a machine can actually create an original, copyrightable work without human intervention is controversial. For now, in most jurisdictions, there is still the presumption that despite the use of machine learning, software, and technology, there is still a human hand behind original works. For example, a couple of years ago a Chinese court addressed this issue, finding that an automated article written by a program called Dreamwriter, created by Tencent, which had been copied and published without permission by another Chinese company, Yinxun, was nevertheless subject to copyright protection because it met the originality test through the involvement of a creative group of editors. These people performed a number of functions to direct the program, such as arranging the data input and format, selecting templates for the structure of the article, and training the algorithm model. In other words, there was sufficient human authorship—and thus there was copyright, and infringement of copyright.
Nonetheless, at times it is a fine line when it comes to assessing how much human creativity has gone into a machine-enabled work. Carlisle provides the example of Dmitry Evgrafov, the writer of computer code for an app called “Endel”. Warner Music signed a deal to distribute 600 tracks created by Endel. What is significant is Evgrafov’s claim that the music was produced with a few clicks of a mouse, with minimal human involvement outside of chopping up the audio and mastering it for streaming. There are machines that allow people to compose music (even though these people don’t even have the ability to read or write music), by manipulating various musical elements, such as rhythm, harmonics, etc. to blend them into a “new” piece of music. How much human creativity is involved? Not much, so is the end product a creation of the software, or is it a human creation enabled by the software?
If AI is a current test case of human authorship, one cannot help but also reference the role of animals arising from the infamous Monkey Selfie case, about which I and others have written. (The Monkey Selfie Case: Applying the Common Sense Test”) and (“The Monkey Selfie Case: Will It Have Broader Repercussions for AI and Copyright?). In this case, the issue revolved around whether wildlife photographer David Slater was the author of the monkey’s selfie because he had not himself clicked the shutter, despite having artistically staged and made technically possible the taking of the photo. Although that was the question, the judge in California ended up ruling whether the copyright belonged to “Naruto”, the macaque that was the subject of the photo, rather than whether the work was Slater’s or in the public domain.
Naruto’s claim was a publicity stunt by PETA, the Society for the Ethical Treatment of Animals, but it ended up financially ruining Slater. Of course, Naruto’s claim (presented on behalf of the animal by PETA as a legal “next friend”) was dismissed, but there were some interesting moments. During the trial there were questions raised as to whether Naruto’s “children and grandchildren” could be handed the copyright, (and how they would be identified) were it to be awarded, and whether Naruto was obliged to send letters to the other monkeys in the group to let them know about the court case, since some of them had also been inadvertently included in photos snapped by Naruto. Just as with the Creativity Machine, it was apparent that a non-human could potentially create something that might qualify as a work, but that an inanimate creator could not exercise any of the other attributes necessary to establish copyright. If a reason is needed to deny copyright protection to works not produced by humans, this is a good one, but still does not finally decide the question as to whether, inevitably, there is always a human creator behind the non-human.
Think of elephants, those well trained, hard-working creatures that not only allow us to ride them (sometimes), but which have been harnessed for centuries for logging, waging war and other human-directed endeavours. Give an elephant a paint brush, a couple of buckets of paint, and a canvas and you might get something that is not worse than some other forms of modern art. If it qualified as an original work, I would award the copyright to the human who created the framework for all this to happen. It is no different than using other means to create art; automatic spray guns, nude bodies smeared in paint, and so on. In this case, the instrument just happens to be an elephant rather than a machine. But someone created the concept and made it happen. A human.
Those who argue that the definition of author or creator should be expanded beyond humans point to the lack of specificity in the laws of most countries when it comes to defining who is an author. The Berne Convention, the cornerstone of international copyright protection, is a case in point. First drafted in the 1880’s it protects “authors”, but nowhere does it define the term. Rather, it protects “every production in the literary, scientific and artistic domain, whatever the mode or form of its expression”. The drafters of the convention could not conceive of a literary, scientific or artistic work that was not the result of human endeavour. So, human creation is implicit. In US law, the Copyright Act protects “original works of authorship”, but without providing a definition of what an original work is. However, the USCO Review Board in Thaler’s case noted that “Courts interpreting the Copyright Act, including the Supreme Court, have uniformly limited copyright protection to creations of human authors”, and cited various references where courts have referred to authors as humans or “a man”, presumably in the sense of “mankind” rather than a gender-specific reference.
While it is generally accepted in most jurisdictions that human creation is an essential ingredient to claim copyright protection, there are some exceptions. Earlier I referred to the United Kingdom. Under the Copyright, Designs and Patents Act, 1988, there is a provision (Section 178) that states computer generated works are works “generated by computer in circumstances such that there is no human author of the work”(emphasis added). These works are currently provided with a more limited term of copyright protection (50 years from the date of creation) as opposed to the standard life plus 70 years provided to works made solely by human authors or authors assisted by AI. (The rights-holder for a fully computer-generated work is “the person by whom the arrangements necessary for the creation of the work are undertaken”, even though that person played no creative role.) The UK Intellectual Property Office (UKIPO) is currently reviewing provisions for protection of AI under the Act, with possible changes ranging from a requirement that human creativity must be involved in order to claim copyright protection, (in other words dropping machine-generated copyright entirely), to reducing the protection for computer-generated works to as little as five years, to retaining the status quo.
In re-examining whether computer-generated works should benefit from copyright protection, the UKIPO consultation paper has set out three considerations; the need to (1) encourage innovation in AI technology and promote its use for the public good; (2) preserve the central role of intellectual property in promoting human creativity and innovation (emphasis added); and (3) be based on the best available economic evidence. Arguments put forward for dropping machine-generated copyright include concerns that protecting computer-generated works may promote these works at the expense of human creations and devalue human creativity. (A machine can work 24/7; humans need to sleep and eat). Proponents for reducing the period of protection for computer-generated works argue there is no need to incentivize a computer to create works, yet protection imposes costs on third parties. Finally, those who support copyright for works produced exclusively with AI maintain that this provides additional economic incentive to invest in AI technology. The public consultation closed on January 7 of this year. No doubt we shall have to wait a while to see what proposals the Office brings forward.
As I wrote in an earlier blog (“AI, Ethics and Copyright”), one factor to consider with regard to the use of AI in creating works is ethical judgement, something only a human can provide;
“Despite enablement by machines and the creation of “new” works using AI through compiling, synthesizing, modifying and re-arranging, at the end of the day there will always be a human creator behind the AI who will judge whether and when the AI work is complete. This person will ultimately take a position on any ethical questions that may arise and the output will be subject to human judgement in terms of choices, editing, finalization, and release.”
I believe that humanity and copyright must go hand in hand. The US Copyright Office apparently agrees although Stephen Thaler clearly does not. The Intellectual Property Office of the UK appears unsure.
What do you think?
© Hugh Stephens 2022. All Rights Reserved.