Recently, the United States District Court for the District of Columbia, in line with several decisions of the U.S. Copyright Office’s Review Board, found that human creativity is the sine qua non of copyrightability, refusing to register a work lacking human creative involvement or control. In this way, the U.S. jurisprudence embraces the distinction between works which are computer-aided and computer-assisted. If the computer somehow transcends its role as a tool in such a way that the purported human author cannot claim he created the output, we are dealing with computer-generatedor emergent works. Such apparently creative outputs are not protectable as a matter of the American copyright law – or the law of the European Union (see Blaszczyk). They may, however, be protectable in the United Kingdom (UK), whose legislators provided protection to computer generated works at the expense of statutory consistency and doctrinal coherence.
I have previously argued that protection of works autonomously generated by artificial intelligence (AI) systems, supposedly lacking a human author, is impossible as a matter of United States and European Union law, but also the international framework, and copyright theory (see Blaszczyk). All of copyright’s conceptual building blocks, the idea-expression dichotomy, originality, authorship, and the concept of a protectable work operate in the notation of human creativity (see Gervais). Autonomous creations of AI systems, consequently, fall outside of copyright’s positive ontology, being akin to ideas, facts, or subject-matter predicated by technical considerations, rather than authorial creativity. They simply cannot be.
This thesis is supported by the curious failure of the United Kingdom’s Copyrights, Designs and Patents Act of 1988 (CDPA 1988) to protect “computer-generated works.” While Section 1 of the Act specifies that copyright subsists only in original literary, dramatic, musical or artistic works (hereafter “works”), sections 9(3) and 178 attempted to protect computer-generated works in circumstances where there is “no human author.” Namely, in works which are computer-generated, the author “shall be taken to be the person by whom the arrangements necessary for the creation of the work.”
We can easily see that the UK legislation is self-contradictory. It provides for protection of authorless works, ascribing authorship to the person making the necessary arrangements, but according to Section 1, without authorial originality, no works are protectable. In this way, the contradictions of the statutory text, mirror the contradiction involved in the very concept of “emergent” or “authorless” works. Without a human author, there is no expression of ideas that can be original, and thus no copyrightable work. The concept of computer-generated works is thus logically inconsistent and incoherent with all of copyright’s doctrinal architecture.
As is well known, the Court of Justice of the European Union (CJEU) harmonized both the notion of originality and that of a work, equating them with “author’s own intellectual creation” (see Rosati). The European standard uses the language of creativity, personality, and creative freedom; it is as far from protecting authorless outputs as possible. In this respect, the UK has been a pluralist jurisdiction: the UK courts followed the CJEU standard, sometimes remarking it was binding but different and more restrictive than the old “sweat of the brow” standard; that the two standards were equivalent or interchangeable; or that the binding standard has been reinterpreted and replaced by the author’s own intellectual creation.
Setting aside the complicated constitutional question whether Section 9(3) of the Statute had been enforceable before Brexit, it was clearly contrary to the European standard for originality and protectable works (see Hugenholtz and Quintais). It may well be that in the future, the UK copyright law will depart from a uniform, creativity based language of the European law; re-interpreting, once again, what originality means (see Richard Arnold et al). However, under no plausible interpretation can an authorless work be original – or an expression of an idea, and thus a work, at all.
Copyright has always developed around a human author, with originality, a qualitative and causal concept, reflecting it well. In the most minimalist English definition, originality designated something originating from the author, which he can claim as his own expression (see Rahmatian). The old cases, like Lockean philosophy, spoke of “mental labour” and, importantly, of authorial “judgement”. A good example is found in the sweat of the brow formula of University of London Press, which proclaimed copyright dealt with the “expression of [authorial] thought”, requiring the work to “originate from the author.” Along with the development of international framework, and the interaction with European law, the domestic standard further evolved in a personalized direction (see Gervais).
Indeed, if copyright considers as existing only that which has been expressed, and the contours of a protectable work are marked by a causal link with its author, then computer-generated works are a contradiction in terms.
These doctrinal and theoretical difficulties were also acknowledged in the Intellectual Property Office (IPO) Consultation (see Gervassis and Trapova). Although Section 9(3) undermines the coherence of the doctrine, the IPO decided not to propose changes. The Government response concludes, that there is “no evidence at present that protection for [computer-generated works] is harmful”; though, we may add, there is no evidence it has made any meaningful impact, and what it might be, with only one case citing it, arguably needlessly.
Of course, there are further various normative difficulties with justifying authorless works’ protection (see Mezei). There are also more prosaic ones: by going beyond the international framework, the UK legislature decided to stifle expressive freedom at home, while not gaining reciprocal treatment abroad (see Ricketson). The recent United States’ case law shows that other jurisdictions may not follow suit. In the UK, like in the case of emergent works at large, it is the public domain, or rather all of us, constitutive of the cultural community, who will be harmed (see Craig and Kerr).
To conclude, the UK statutory provisions demonstrate a curious case: they are self-contradictory, practically insignificant, and widely criticized (see Grimmelmann; Ramalho; Aplin et al). The contradictions of the British Statute show well that copyright law simply cannot encompass emergent or computer-generated works coherently. All of copyright’s framework is anthropocentric: much of its shape, including the very concept of a work, owes to Enlightenment romanticism and cognitivism; is built upon authorial expressiveness. This is the language of copyright’s game; and attempts to undermine it, for reasons poststructuralist, post-humanist, or otherwise, not only distort the shape of the common concepts and the integrity of the copyright’s system (but see McCutcheon), but go against the cultural common good (see Frosio). Perhaps, even if there was nothing in nature which differentiates human creators from machines (see Bridy, cf. Frye), we can differentiate ourselves through law and the narratives and myths which surround it. This is, in fact, one of the great beliefs of modern jurisprudence.
This post summarizes and builds upon the main findings of the article Impossibility of “Emergent Works’ Protection in U.S. and EU Copyright Law,” published in the North Carolina Journal of Law & Technology, available at: https://ssrn.com/abstract=4519511.