WHO PUTS THE "ART" IN ARTIFICIAL INTELLIGENCE?
[Authored by Atreya Mathur, an NYU LL.M. Candidate ’21 specializing in Competition, Innovation and Information Law with a focus on Intellectual Property Rights.]
This post examines art created by Artificial Intelligence (AI) as protected under copyright law in India and the United States. It considers such artwork as postmodern/contemporary art that satisfies the originality criteria for copyright protection. Further, the need for authorship/ ownership and protection of the artwork is considered, analyzing the implications of releasing such artistic work created by AI into the public domain.
“Alexa, write a poem for me.” Or “Hey, Siri. Can you finish the design for my art project?” Imagine having the power of commanding software to create content that generally requires intense effort and skill. Does this come under artistic work? Would it be yours? Would it be Alexa’s? Or would it belong to Amazon (the creator of Alexa)? Can you use the work, or are you violating someone else’s or something else’s copyright?
At the outset, it may seem strange and quite frankly impossible for anyone or anything to create artistic works other than a natural person; a human. Art is an intellectual creation of the mind; it requires capability, creativity and potentially even humanity to some extent, so is this a little far-fetched? Maybe. But does that mean that it is impossible? If not, to what extent is it possible? Can such works of art be afforded protection? And to whom is this protection granted-the creator of the artwork or the creator of the Artificial Intelligence (AI)? Copyright law, as of now has left several issues unanswered with the niche and exciting field of AI gaining significant traction over the past few years.
In 2016, an AI program was almost awarded the Nikkei Hoshi Shinichi Literary Award in Japan. In 2018, the Portrait of Edmond Belamy was sold at Christie’s New York auction house for $432,500; the artwork created by an algorithm defined by the algebraic formula appearing as the artist’s signature on the piece. More recently, Warner’s Music signed the world’s first-ever record deal with an AI algorithm to produce 20 albums. AI created by humans is now creating art without human intervention.
PILOT: THE ONE WHERE AI MEETS ART
“The rise of the machines is here, but they do not come as conquerors, they come as creators.”[i]
Artificial intelligence has solved some complex and interesting issues through the employment of directed algorithms designed by humans but has in the process created more intriguing problems that need to be addressed. Whether it’s the iconic robot Sophia who strives to “start a family” or the Instagram model Lil Miquela- a digital influencer stating “I’m not a human but am I still a person?”, AI has given the world a lot to think about.
The machine learning and intelligence behind the technology have allowed lawyers to spot contractual loopholes in legal agreements and has even helped the world by detecting spam emails to keep accounts organized and clutter-free. In more recent times, algorithms are being created allowing computers and machines to create works that are original sans any direct influence by humans with regard to the result; the ultimate creation.[ii]
In the contemporary world where there is a substantial and widespread use of AI by artists, the line between artistic works created by humans and those by machines have become blurry. The result of which has fascinating implications on copyright law that protects works created only by humans. With Artificial Intelligence now generating original stories, music and art, the law needs catching up on issues of copyright ownership.
COPYRIGHT LAW AND OWNERSHIP OF ARTWORK
Copyright refers to an exclusive legal right providing protection to original creations of the human mind and intellect. In India, Section 14 of the Copyright Act 1957, defines "Copyright" as the exclusive rights of an owner to do or authorize the doing of any acts including but not limited to the reproduction, adaption and translation of the work. Further, Section 17 specifically states that the author of the work would be the first owner of the copyright, but in the case, a work is created under a contract for consideration and is dependent on the instruction of employer, then the owner of the work would be the employer. A court in India held that “In the context of question papers for an examination, that the author of the examination paper is a person who has compiled the questions; the person who does this compiling, is a natural person, a human being, and not an artificial person[iii]; the Central Board of Secondary Education is not a natural person, and it would be entitled to claim copyright in the examination papers only if it establishes and proves that it has engaged persons specifically for purposes of preparation of compilation, known as question papers, with a contract that copyright therein will vest in Central Board of Secondary Education.”[iv] It may be clearly inferred that in India, as evidenced by such cases, for the purpose of copyright, only natural persons must be provided as the author of the work. [v]
While most computer-generated artworks generally rely heavily on the programmer who creates the AI, in recent times AI has developed to the extent that it can generate results without any interference by humans.
UNDERSTANDING ARTWORK CREATED BY "NON-HUMANS"
The fundamental question before addressing AI-created art is whether copyright can belong to anyone other than a human being. The Naruto Case in the US throws some light in understanding the matter and answers this question.
David Slater, a British (and very much human) wildlife photographer, set a camera up on the island of Sulawesi and had left the camera unattended. A handsome and rather curious young gentleman (a monkey) named Naruto clicked the button while looking at the camera, capturing selfies showing off his photogenic side. Following this, the human photographer published a book featuring the selfie and other pictures that had been taken by Naruto.
In response to this, the People for the Ethical Treatment of Animals (PETA) filed a complaint against him and the publisher, representing the monkey. Naruto had argued that he had: “the right to own and benefit from the copyright in the Monkey Selfies in the same manner and to the same extent as any other author.” Further, it was argued that “While the claim of authorship by species other than homo sapiens may be novel, “authorship” under the Copyright Act, 17 US.C. § 101 et seq., is sufficiently broad so as to permit the protection of the law to extend to any original work, including those created by Naruto.” However, PETA lost the case as animals have no standing under the Copyright Act in the United States. Animals do not have the locus standi to file a copyright infringement suit or any other lawsuit. The US. Copyright Office stated that they “will refuse to register a claim if it determines that a human being did not create the work.” The office also said that it would exclude works “produced by machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.”
To put it simply, the Ninth Circuit dismissed the suit as Naruto could not own the copyright to the photos. What is imperative to note in this case, is the essential question raised by this decision-whether computer-generated art would be copyrightable and in such a situation with whom does ownership lie?
In order to understand the difference between art created by a non-human such as an animal or monkey in the case of Naruto, it becomes crucial to know how AI creates art in the first place.
Barrat, who is a self-taught programmer cum AI artist, states that AI artists use two neural networks that consist of algorithms called a generator and a discriminator to create AI art.[vi] To illustrate, Barrat would feed the generator paintings; the generator would then create rules based on those paintings in order to produce its own version of them.[vii]The discriminator then plays its role by looking at both the real paintings and the AI versions to determine which painting is “real.” The generator would then try to trick the discriminator into believing that its new paintings are “real.”[viii] The resulting AI work could be partially or substantially different from the base underlying artwork adding to the originality component necessary for the artwork to be given copyright protection-but this is case dependant. It should be noted that the work of AI may qualify as original work, but it is highly debated whether the work so created uses “skill and judgement” by the AI which is required in India as a component for artwork to be protected as original work. However, since the creator of the AI generally compiles the work, this may potentially qualify as the required skill and judgment necessary for an original creation through attribution of the artwork to the programmer.
A distinction must also be made between art created by AI and the computer software used to create the work itself as some aspects of a computer program can be protected by copyright. So, if the tool used to create a work is protected by copyright, does that mean that the work that is ultimately created is also protected as computer software? The answer is no; computer software is separate and distinct from the work it creates and cannot be protected as one. In the US, it is specified that “ownership of the copyright in a work is distinct from ownership of any material object that may be used to create that work. The fact that the author used a computer to write an article, short story, or other nondramatic literary work does not mean that the work is a computer program.”[ix]
Therefore, the protection of such work created by AI is substantially different from the Naruto case, as a natural person created an algorithm that created its own work. (which is not protected as computer software.) Now, the questions are: 1) Would this work fall under the definition of art? 2) would it come under copyright protection? And moreover, 3) Who finally owns it?
DO ARTWORKS CREATED BY AI FALL UNDER COPYRIGHT PROTECTION?
Defining art is a challenging task due to the inherent subjectiveness of the nature and perception of art. According to the Oxford dictionary, art is defined as: “The expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power.” [x] While this definition states “human creative skills” is required, AI-created art would mostly fall under contemporary artwork.
Contemporary or postmodern art defines art in an unconventional way so as to come to a definite understanding of whether the copyright laws in place for art could be considered as workable or not upon the examination of the new avenues of contemporary art.[xi] It is argued that postmodern artists do not necessarily take into consideration the economic rights associated with the artwork and they would therefore not require the incentives provided by copyright laws that would inspire such artists to question the conventional or traditional definition of art. This premise would favour AI created Art as AI does not need any sort of motivation to create the artwork- it is art created for art’s sake not generally for monetary incentives. Therefore, as long as it is an original work of authorship, is fixed in a tangible medium and has a certain amount of creativity (satisfying copyright criteria) it would be protected under copyright law as an artistic work. Through this, it may be concluded that art created by AI comes under the definition of artistic work as contemporary or postmodern art, which would fall under copyright protection.
The requirement of novelty or originality in a work generated by AI is not usually disputed. In China, a court[xii]observed that an article that was written by an AI was considered as “an original work.” Despite this, the court stated that the work could not be protected under the copyright laws as the work was not created by a natural person. In the US, the Copyright Office follows their traditional policy of rejecting claims that are made for works not authored by humans; the policy which is considered to be poorly codified.[xiii] However, there does not seem to be an actual requirement for human authorship in the US Copyright Act itself that could potentially extend the scope of authorship. The courts so far have always only assumed the concept of authorship to be a human phenomenon, which is similar to the view of courts in India.
So while it may, therefore, be presumed that artwork created by AI comes under contemporary art (provided it meets the originality criteria) that could be protected by copyright, ownership makes copyright protection more complicated-as without authorship and ownership, work remains unprotected and is released into the public domain.
SO WAIT, WHO OWNS THE ART?
The apparent transition of AI technology from being merely generative to being creative renders for more meaningful involvement of the ‘intelligent agents’ in the law.[xiv] The law therein poses a challenge to ownership and enforcement of rights in a work that has been created by artificial intelligence.[xv]
The artwork is a subject matter of copyright and is protected, provided it is original and satisfies copyright criteria. Rights, including the right to communicate to the public and to reproduce or adapt, are granted to the artist. The law does not explicitly state if a non-natural person can be considered as an artist, but through cases such as that of Naruto, it has been established that a non-natural person cannot hold the copyright.
Thus, a painting created by a machine cannot be an author that owns the copyright, but this does not restrict AI artists like Barrat from claiming an element of authorship and ownership in the resulting work. Since an artist’s initial input is necessary for the AI to create work, the artist arguably exercises some degree of originality in the selection of the underlying works that form a basis for the AI created work. Therefore, where work is created by AI with human interference, the authorship can be attributed to the human. This essentially means that the extent of input by the initial artist would allow for such artist to claim authorship for the artwork as a derivative work. In some cases, it may even be considered where work is created by the AI without any human interference; ownership may be claimed by the copyright owner of the AI (the person who holds copyright over the AI software).
Some countries do protect such works and grant the initial artists copyright over artistic creation. For example, the UK allows copyright protection to the person that makes arrangements for the computer to create the work.[xvi]
THE FAIR USE AND INFRINGEMENT CONFLICT
Another issue that may arise when the initial artist creates an AI to produce art is the fair use and infringement conflict. This is predominant in AI-created artworks as the input for generating the art is reliant on already existing works of art. The computer program is fed copies of existing art to try to recreate the work, so the line between fair use and infringement is very thin that too if there is a line at all. However, this may not always be the case as works can differ and be considered original on a case to case basis given the arbitrariness of the artwork created. Prior appropriation art cases suggest that fair use may be viewed as a defence to infringement if the result is sufficiently transformative and can be seen as a deviation from the already existing artwork.
AI artists should therein both document the creative process when selecting and inputting the underlying art and additionally consider evaluating the final AI artwork to determine whether it is sufficiently transformative before releasing it to the public to reduce potential infringement claims.
THE FUTURE: WHERE DO WE GO FROM HERE?
The future is AI, and the future is here. This is merely the beginning of works being created by AI- whether it is art or music or poetry. The most effective solution would be to follow a system similar to the UK, where copyright is granted to the person who made the AI. Additionally, this would also ensure companies keep investing in the technology, knowing that the business would hold copyright receive benefits associated with the copyright granted. Alternatively, another solution to the ownership problem may be allowing companies which are recognized as legal people to be awarded copyright for works they don’t directly create, through effectively commissioning the AI to produce work for it allowing for authorship to belong to such human or company. If an AI-generated artwork is not protected by copyright, it belongs to the public domain, and anyone can freely use it, make copies of it, distribute it, or use it for commercial purposes and sell it to others. This would prove to be futile as the overall impact of being in the public domain may result in discouragement from developing AI systems, or working with them to create original artworks. Making it free may have several advantages that may prove useful in a relatively futuristic setup (such as in Amazon Prime’s “Upload” set in 2033- where making AI available to the public would probably be more beneficial for access, reducing costs and for more robust development of glitching software!). However, since this is only the beginning of the age of developing such high tech software, AI-created artwork in the public domain may slow down technological progress and advancement, and reduce potential investment in the technology.
While the UK model may be the most effective solution, it may not be the most correct or apt one in certain circumstances. AI is now creating work that significantly outperforms the work created by the original artist. Google’s AI managed to create its own AI that proved to be more efficient than what was created by humans. In the same way, if Alexa or Siri write a poem or create an online design, ethically should the copyright not lie with the AI itself and not with Amazon or Apple or the person who commands such art with no remuneration provided? To draw a parallel, in the Netflix series “Black Mirror”, one of the many intriguing situations the show brought out were the ethics and moral principles in killing AI who believed itself to have a consciousness. If Kantian philosophy is to be applied to such a situation, ultimately the morality is not just in the act of killing but whether one believes the AI to be human or not. Maybe this could be compared to the legal status granted to a company in the sense that dissolving a company (which is legal) in essence is killing the company, however, the difference is that a human coded the AI to believe the AI is real resulting in such AI developing its own consciousness. This may impose moral responsibility on the creator of the AI. Considering these hypothetical (and at this point seemingly impossible) scenarios, in the same manner, only if AI is granted the status of a legal entity, and considered as an entity capable of ownership (in the eyes of the creator, the public and the law) would copyright over the artistic work be awarded to the AI.
Therein, this creates not only a legal question but also begs for a philosophical and moral inquiry into whether AI should be given the status of a legal person due to accompanying legal and moral rights that the AI would potentially possess. This, however, is an ongoing debate with several sub-issues that have their own set of legal ramifications.
WHAT HAPPENS IF AUTHORSHIP OR LEGAL STATUS IS GRANTED TO AI?
The special rights of the author including, moral rights, right to paternity and right to integrity, may be disputed. While AI-created art could fall under contemporary art devoid of the need for monetary incentives, moral rights may still be debated. If an AI is recognized as the author of a work, then these rights would at the outset become redundant, as the rights generally have the notion of human feeling and emotion accompanying the work. This would once again require a more philosophical outlook on the matter.[xvii]
With regard to the copyright law of India, the author of the work has a right to claim royalty which cannot be waived. In the case of an AI being the author of a work, the determinability of such fee and its disbursement becomes difficult.[xviii]
For any work created by AI, accountability of AI over any creation will also prove to be challenging to enforce. Section 51 of the Indian Copyright Act 1957 states copyright can only be infringed by a "person". Infringement caused by AI becomes a critical issue in the case of placing liability. If AI is granted legal status, enforcement of such provisions will become difficult even though the liability placement may become more manageable. In the current scenario, if there is infringement by AI, the author of the AI would most likely hold liability for infringement.
It may be concluded that the current copyright law of India as of now is not well equipped or exhaustive enough to grant rights to AI for the creation of work. Through various cases, the courts have also focused on the need for human interference for copyright protection. In addition, there is minimal scope to accept AI as a separate legal entity, especially in comparison to the US.
UNCERTAINITY: TIME WE GET USED TO IT?
Regardless, the issue can only be resolved on a case-to-case basis that would consider the level of AI involvement. If the AI, such as in the case of google has a substantial role and can be regarded as the author of the work, then under current law the artistic work would not be eligible for copyright protection at all. If the machine is granted the status of a legal person, copyright could lie with the AI. If not, then to the extent of human input by the initial artist, such artist could be deemed the author and receive copyright protection for their creation of the work.[xix] However, this work may still only be considered a derivative work.
In conclusion, to ensure there is advancement of innovation, increased commercial gains, investment in and development of such technology (in its nascent stage), such works of art should be provided authorship and must be copyright protected-either as original work of an AI (subject to being considered as a legal entity) or as a derivative work of the initial artist rather than allowing the work to fall into the public domain.
“By far, the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.”
- Eliezer Yudkowsky
The future in terms of the legal protection of AI-generated works is still dubious but exciting, as it is close to impossible to understand the full extent to which AI can potentially be used and will be used for the creation of artistic works. Until then, we can only imagine the legal implications of what might be in times to come.
[i] Andres Guadamuz, “Artificial Intelligence and Copyright”, WIPO Magazine (Oct. 2017) available at https://www.wipo.int/wipo_magazine/en/2017/05/article_0003.html. [ii] Courtney White, Rita Matulionyte, “Artificial Intelligence: Painting The Bigger Picture For Copyright Ownership” (Dec. 2019) available at https://ssrn.com/abstract=3498673 or http://dx.doi.org/10.2139/ssrn.3498673. [iii] Rupendra Kashyap v. Jiwan Publishing House Pvt. Ltd. 1994 (28) DRJ 286. [iv] Lucy Rana, Meril Mathew, “Artificial Intelligence and Copyright- the Authorship (Dec. 2019) available at https://www.mondaq.com/india/copyright/876800/artificial-intelligence-and-copyright-the-authorship [v] Id. [vi] Joe Dworetzky, “Q&A: Robbie Barrat on training neural networks to create art,” The Stanford Daily (Jun. 2018) available at https://www.stanforddaily.com/2018/06/12/qa-robbie-barrat-on-training-neural-networks-to-create-art/. [vii] Id. [viii] Id. [ix] Section 721.6 of the Compendium of US Copyright Office Practices available at https://www.copyright.gov/comp3/docs/compendium.pdf. [x] Oxford English Dictionary. Oxford University Press, 2000. [Electronic resource]. [xi] Maarten Dorman, Sherry Marx, “Art in Progress: A Philosophical Response to the End of Avant-Garde” (2003). [xii] Beijing Feilin Law Firm v. Baidu Corporation, No 239 , Civil First Instance, Beijing Internet Court, 25 April 2019. [xiii] Robert David Hart, “If an AI create a work of art, who owns the rights to it?” (Aug.2017) available at https://qz.com/1054039/google-deepdream-art-if-an-ai-creates-a-work-of-art-who-owns-the-rights-to-it/ [xiv] Subhash Bhutoria, “From Mona Lisa to Machine Learning - Artificial Intelligence and Art” (May 2020) available at https://www.artlawindia.com/post/ai-and-art [xv] Id. [xvi] Andres Guadamuz, “Artificial Intelligence and Copyright”, WIPO Magazine (Oct. 2017) available at https://www.wipo.int/wipo_magazine/en/2017/05/article_0003.html. [xvii] Lucy Rana, Meril Mathew, “Artificial Intelligence and Copyright- the Authorship (Dec. 2019) available at https://www.mondaq.com/india/copyright/876800/artificial-intelligence-and-copyright-the-authorship [xviii] Id. [xix] Stan Gibson, Jessica Newman, “Blurring the Lines: When AI Creates Art Is It Copyrightable?” available at https://patentlaw.jmbm.com/2020/05/blurring-the-lines-when-ai-creates-art-is-it-copyrightable.html