E-David is a prodigy artist robot equipped with an arm, five brushes, a camera and an optimisation system via visual feedback. Developed by the Department of Computer and Information Science at the University of Konstanz in Germany, E-David creates original artwork by autonomously taking pictures with its camera, using a complex visual optimisation algorithm and accordingly creating original paintings from these photographs.(1) Such original artwork, if created by a human being, would certainly be capable of copyright protection.
Over the last decade AI systems and machines have been passing the so-called 'Turing Test', which determines whether a computer system can play the 'imitation game' and exhibit intelligent behaviour equivalent to or indistinguishable from that of a human. A computer is deemed to pass the Turing test if a remote human evaluator is unable to distinguish whether the output is machine or human created.
In recent years there has been an increase in creative AI systems passing the Turing Test to produce original copyrightable works. For instance, the Iamus software program creates musical works that have been performed by the London Symphony Orchestra.(2) Similarly, in 2011, Zackary Scholl created poetry-writing software which produced a poem that was accepted for publication in a literary journal and the editors did not realise that it had been written by a computer program.(3)
AI systems are evolving rapidly and their capacity to be creative, autonomous, rational and self-learning are blurring the lines between original works, which are products of human intellect capable of copyright protection, and mere computer-generated works. As such, the AI landscape is challenging conventional copyright laws in India and internationally and raising a number of legal implications and ambiguities regarding ownership, authorship and accountability in AI-generated works.
For copyright to subsist in literary, dramatic, musical and artistic works, Section 13 of the Copyright Act 1957 specifies that it must be original. As the act does not define the term 'original', in Eastern Book Co v D B Modak the Supreme Court adopted the 'modicum of creativity' standard for determining whether a work will be afforded copyright protection.(4)
According to this standard, original copyrightable work must meet the "minimum requirement of creativity" and should not be merely an outcome of skill and labour. As the threshold requirement of creativity is not particularly high, an AI-generated work may meet this originality standard and be copyrightable.
However, pursuant to meeting the modicum of creativity standard, the second essential question under the Copyright Act is deciding the author of a work. Section 2(d) of the act defines the author of "any literary, dramatic, musical or artistic work which is computer-generated" to be "the person who causes the work to be created". This definition is aligned with UK copyright law.
As AI-generated works fall under the category of computer-generated works, the author of an AI-generated work will not be the AI system, but rather the person who caused the work to be created. By excluding artificial persons, this definition implies that only natural persons can be protected as authors under the Copyright Act. In fact, attributing authorship of computer-generated works to natural persons is confirmed in the Practice and Procedure Manual (2018),(5) issued by the Copyright Office, which states that in the application for copyright registration for literary works, only "the details of the person(s), who has actually created the work i.e. only natural person (human being), should be provided".
The Delhi High Court's judgment in Amarnath Sehgal v Union of India is useful to examine the essence of authorship under copyright law in India.(6) In this judgment the court recognised "authors moral rights" pursuant to Section 57 of the Copyright Act (ie, the right of paternity, the right to maintain purity or integrity in a work and the right of retraction). These rights, the court observed "originate from the fact that the creative individual is uniquely invested with the power and mystique of original genius, creating a privileged relationship between a creative author and his work". Therefore, in cases where the work is AI generated, the AI system's recognition of this privileged relationship between work and author will be difficult to establish. An AI system may be unable to ascertain the 'morality' behind the underlying reputation and honour associated with the work and the author.
Therefore, while an AI-generated work may be deemed 'original' in the current technological landscape and given the extant legal framework, the AI system may not be attributed authorship thereof.
As AI systems are not natural persons and AI-generated works are considered computer generated under Section 2(d) of the Copyright Act, an ambiguity exists regarding the identification of the "person who cause[d] the work to be created". Is it the person who programmed or created the AI system, the owners of the AI system or companies and financial investors in the AI sector or is it the end user who uses the AI system to generate a certain output? The lack of clarity and the complexities involved in determining the author of an AI-generated work in turn make it difficult to determine the 'first owner' of a copyright under Section 17 of the Copyright Act, which is typically the author of a work, subject to certain statutory exceptions.(7)
A related issue arises from the fact that, under Indian copyright law, in certain situations copyright ownership may be granted to non-natural, legal or juristic persons (eg, companies, organisations or the government). Therefore, if future AI systems are declared legal or juristic persons, they could be granted copyright ownership in certain situations; however, this would generate issues relating to copyright transferability and the financial and commercial aspects of copyright ownership.
In general, AI systems are not considered natural persons (thereby removing the issue of affording AI systems copyright authorship), but the lines appear to be blurring with regard to the recognition of AI systems as legal persons.
The Indian courts have yet to address these intricate issues surrounding AI-generated works and copyright authorship and ownership.(8)
As AI systems develop capabilities traditionally attributed to humans, such as creativity and autonomy, preconceived notions about human intelligence and creations of the mind are challenged, which in turn puts pressure on existing legal frameworks to evolve.
With the reduction in human intervention in AI systems and AI-generated works, policymakers globally may eventually have to create systems and codes which consider the moral, commercial and accountability aspects of copyright protection in AI-generated works as well. It would be interesting to see how the law evolves to protect and encourage AI developers and users on the one hand and AI systems and their potential juristic personality on the other.
(1) Shlomit Yanisky-Ravid, "Generating Rembrandt: Artificial Intelligence, Copyright, and Accountability in the 3A Era – The Human-Like Authors are Already Here – A New Model", Michigan State Law Review, 659 (2017).
(3) Brian Merchant, The Poem that Passed the Thring Test, VICE Motherboard (Accessed 5 February 2015). Available here.
(5) Practice and Procedure Manual (2018), Copyright Office, Goverment of India. See here.
(7) The Chinese courts have ruled that AI-generated works are entitled to copyright protection and have attributed ownership to AI tool creators. See "AI produced work is also entitled to copyright; rules a Chinese court", TechGig. Available here.