Traditional Theories of Ownership in IP and What Dreamwriter Means for AI-Generated Works
As technology grows increasingly more sophisticated, it has transformed from a passive tool that requires human guidance to essentially the sole author of its own artistic output. The rise of AI-generated creativity raises important questions about authorship as it relates to copyright, and whether the current US copyright system is adequate to address the ways in which technological innovations challenge traditional theories of ownership.
U.S. copyright law grants copyright in “original works of authorship.”[1] Notably, the Constitution empowers Congress to “secure authors with exclusive rights to their writings.”[2] While authorship is not expressly defined in the Copyright Act itself, courts have defined its parameters through case law. And while the U.S. Copyright Office lacked statutory backing for their policy of not allowing non-human authors, federal courts have been consistent in their interpretations of the law. Courts have overwhelmingly analyzed conflicts in ownership rights from the lens of authorship. In Community for Creative Non-Violence v. Reid, the Supreme Court defined the author as the party who creates a work, clarifying that it must be a “person who translates an idea into a fixed, tangible expression…”[3] In Uranti Foundation v. Maaherra, the Ninth Circuit also defined authorship as the “first human beings who compiled, selected, coordinated, and arranged [the work].”[4]
The motivation to define authorship through personhood may be partly explained by the courts’ emphasis on the idea of inspiration, which is conceptualized as “uniquely human.”[5] In Burrow-Giles Lithographic Co. v. Sarony, the court defined an author as follows: “he to whom anything owes its origin; originator; maker; one who completes a work of science or literature.”[6] The court in Bleistein v. Donaldson Lithographing Co., held that the author’s unique personality and reaction to nature is the essence of a copyrightable work.[7]
The U.S. Copyright Office further clarified that copyright protections apply only to human authors, following Naruto v. Slater. In this case, the main issue involved whether a monkey could assert statutory authority under the U.S. Copyright Act to sue for copyright infringement.[8] The Court of Appeals upheld the lower court’s decision that non-human animals lacked statutory standing under the Act, relying on the common-law concept of authorship. Specifically, the Court held that “if an act of Congress does not ‘plainly state’ that animals have statutory standing, then the answer is clear, and it can be concluded that they, in fact, do not.”[9]
The U.S. Copyright Office enshrined this holding by affirming that the Copyright Office will only register an original work authored by a human being under section 306 of The Human Authorship Requirement in the Compendium.[10] The Compendium reinforces this requirement in section 313.2, clarifying that works which lack human authorship are ineligible for copyright registration, specifically mentioning “a photograph taken by a monkey” as an example.[11] The effort to exclude non-human beings from copyright protections has been extended to artificial intelligence: the Compendium plainly states that the Office will not register works created by a “machine or mere mechanical process” that operates autonomously without the creative participation or intervention of a human author.[12] Moreover, even if a human and an AI were to produce an identical work, “applying the context, the history, and the legal analysis of the Monkey Selfie case, autonomously created AI works would be rejected by the Copyright Office…”[13]
These court decisions support a notion of creativity as something solely capable of being exercised by human beings. By that reasoning, a work of literature, a painting, or a sculpture, is creative precisely because it reflects the lived experience of its human author. The very uniqueness that seems to characterize a work of art is tied to the uniqueness of the artist and the particularity with which they experience the world.
Implicit in the US copyright system is not only the idea of human authorship, but of individual or solitary genius. The image of the lone artist is rife throughout films and literature, and often comes hand in hand with notions of creativity. This has important implications for the theories of ownership recognized by the intellectual property regime.
Individual ownership begins with the notion of one person holding one exclusive IP right and reflects a Western understanding of the right to property.[14] This understanding of ownership as a private right stems from a system that seeks to both encourage and reward individual success in innovation, creativity, and investment.[15] In theory and in practice, the notion of individual ownership entails splitting up the world into two categories. One category includes the owner and recognizes their exclusive right to a particular subject matter; the other encompasses everyone else, who is necessarily excluded from using that subject matter because they lack the corresponding right in that subject matter.[16]
In contrast, traditional knowledge (“TK”) systems predating contact with Europeans have existed in many parts of the world for thousands of years.[17] Unlike the Western IP system, traditional knowledge systems and the rules around which they are organized vary widely and lack international consensus.[18] While all traditional knowledge systems are characterized by their ties to a cultural tradition tied to a particular community, the boundaries of what should be protected and why are treated differently across national legislations.[19] Notably, not all traditional knowledge originates from Indigenous populations. Forms of knowledge like ancient Chinese medicine, Caribbean steel drum making and music, ancient Belgian weaving, and Swiss yodeling have all been acknowledged as forms of traditional knowledge. However, the vast majority traditional knowledge does originate from Indigenous peoples.[20] Importantly, the type of Indigenous knowledge that falls under this categorization does not include contemporary Indigenous knowledge and knowledge stemming from a combination of traditional and contemporary knowledge.[21]
Included in WIPO’s working definition of traditional knowledge is a characterization of TK as “know-how, skills and practices that are developed, sustained and passed on from generation to generation within a community, often forming part of its cultural or spiritual identity.”[22] Equally significant are the ways in which traditional knowledge systems have been devalued by Eurocentric views. By their very nature, traditional knowledge systems depart from the individualism that pervades the Western IP system. Indigenous customary law “has an inextricably communal nature.”[23] And the social structures that preserve and operate these laws through successive generations “are deeply rooted in the traditional territories of Indigenous peoples.”[24] In this way, the laws reflect an understanding of these communities and the land and environment with which they interact as inextricably linked. And while neither the common law nor international treaties treat these laws with the same deference as other sources of law, they remain an important perspective in the discussion of ownership in the context of copyright law.[25]
In the same way that traditional knowledge systems recognize an underlying and essential relationship between individuals and community, community and environment, AI-generated creativity compels a restructuring and a rethinking of ownership and authorship at a time when creative output is no longer solely attributable to a human source.
Different countries and their legal systems have taken diverse views on copyright protections for works generated by AI. China, a leader in the use of AI in journalism, is one of the first countries in the world to set a court case precedent that protects copyrights for AI-generated works. In 2019, a Chinese court granted copyright protection for an article written by Dreamwriter, a news-writing bot, claiming that the human intellectual activities of the AI program’s creators extend to the works written by the software.[26] Developed by Tencent Technology, Dreamwriter is an algorithm-based intelligent writing assistance system.[27] The Court held that the article generated by the Dreamwriter software was a written work protected by Chinese copyright law, and that the Plaintiff owns the copyright.[28] Shenzhen Tencent v. Shanghai Yingxun raises important legal issues regarding AI in terms of copyright, including whether an AI-generated work can become a work protected by copyright law, and assignment of copyright ownership in AI-generated works.
Significantly, in Shenzhen Tencent v. Shanghai Yingxun, the Court found that while the article generated by Dreamwriter was a written work, the article was created by the creative team of the plaintiff Shenzhen Tencent using Dreamwriter software.[29] In doing so, it deferred to the theory that a work must be the result of an author’s intellectual creation. Because the article incorporated the selections of the creative team in data input, template and corpus style choices, it could not be solely attributed to the AI.[30] The involvement of human actors and their intellectual ideas reinforced the protection of the work under the copyright law.
In considering whether autonomously generated works by AI can be granted copyright, the AI’s capacity to form new, autonomously generated algorithms apart from those pre-programmed by humans and the products resulting from these autonomously generated algorithms “could be called autonomously generated products of AI.”[31] However, according to the current judicial practice of the Chinese courts, cases addressing the issue of copyright protections for autonomously generated products of AI have yet to arise.[32] And some courts continue to note that “The creation of a natural person should still be a necessary condition for a work to be copyrighted under the Copyright Law.”[33] Moreover, the court in Shenzhen Tencent found that because the article in question was the result of the intelligence and collective efforts of multiple teams presided over by the Plaintiff, the copyright of the completed work belonged to the AI developer.[34] Further, because there are no clear answers relating to the copyrightability of the autonomously generated products of AI, the issue of copyright ownership remains premature.
But part of the issue with a decision like Dreamwriter is that it places copyright back in the AI developer or the AI user—essentially a human actor. In doing so, it can perpetuate a power dynamic that takes the innovation of the AI and transforms it into a wealth-generating asset that disproportionately favors tech companies with the infrastructure to both build and capitalize on AI-generated works by dominating copyright ownership in these works.[35]
In this way, despite the considerable advancements of AI and its capacity to autonomously generate creative works, it once again becomes a tool that can be abused and misappropriated by human actors. In doing so, it continues to perpetuate the very power structures that alienate rich sources of traditional knowledge from the IP regime and fails to provide valuable protections. Perhaps a way around this hurdle is to surrender all AI-generated creativity that is truly autonomous to the public domain. It certainly would align more with the goals and purposes of the copyright system. AI can be recognized as authors capable of generating creative works, but the benefits of these innovations can be shared openly with the public, reinforcing the very values on which the copyright system was built: to encourage creativity and innovation and thereby enrich society.
Footnotes
