Content uploaded by Virendra Ahuja
Author content
All content in this area was uploaded by Virendra Ahuja on Jul 18, 2023
Content may be subject to copyright.
ILI Law Review Winter Issue 2020
270
ARTIFICIAL INTELLIGENCE AND COPYRIGHT: ISSUES AND CHALLENGES
V. K. Ahuja
ABSTRACT
The increasing role of Artificial Intelligence in the area of medical science, transportation, aviation, space,
education, entertainment (music, art, games, and films), industry, and many other sectors has transformed our day to
day lives. The area of Intellectual Property Rights is not an exception. The role of AI in creativity and innovation
has been recognized worldwide. AI has a significant role to play more specifically in copyright, patents, designs, and
trade secrets among various types of IPRs. AI can inter alia, compose music, write blogs, novels, poetry, generate
paintings and drawings. The distinction however, has to be made between the works created by a person with the
assistance of AI and the one created by AI itself without any human intervention.
AI has created serious issues and posed challenges in the areas of intellectual property rights more particularly in
copyright law. The present article discusses how AI is important in producing creative works such as arts, music,
poem, novels among other things. The article will also discuss the issues of authorship and “deep fakes” in the work
produced by the AI autonomously. The article discusses the legal position in several countries and deals
comprehensively with several models of authorship in AI-generated works. The discussions being made in this
regard at the WIPO have also been discussed in the article.
I. Introduction
II. Artificial Intelligence
III. Artificial Intelligence and Copyright
IV. Artificial Intelligence and Data Protection
V. Conclusion
I. INTRODUCTION
ARTIFICIAL INTELLIGENCE (AI) has assumed significant importance in the contemporary
times as its use has become indispensable in most of the technological applications. AI has
transformed our lives by entering into various sectors such as heath, transportation, aviation,
space, education, entertainment industry (music, art, games, films) and others. A tendency has
Professor, Faculty of Law and Joint Director, Delhi School of Public Policy and Governance (Institution of
Eminence), University of Delhi, Delhi-110007.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
271
been developed in all countries to automate most of the activities and minimize human
intervention to ensure efficiency and rule out errors.1 Prof Stephen Hawking once said that “the
development of full artificial intelligence could spell the end of the human race”. He further said
that “it would take off on its own, and re-design itself at an ever increasing rate” and “humans,
who are limited by slow biological evolution, couldn’t compete, and would be superseded”.2
It is noteworthy that the Google AI system has become advanced to the extent that it has created
a child of its own. The child AI is being trained by the parent AI to “such a high level that it
outperforms every other human-built AI system”. The performance of child AI is evaluated by
the parent AI which acts as a controller. The information so received is used to improve the child
AI’s performance. This process is repeated thousands of times to make the child AI more
effective and advanced.3
The increasing role of AI in the area of creativity and innovation has been recognized worldwide.
Recently, the OpenAI, an artificial intelligence lab in the United States unveiled a new AI system
called GPT-3 which spent several months “learning the ins and outs of natural language by
analyzing thousands of digital books, the length and breadth of Wikipedia, and nearly a trillion
words posted to blogs, social media and the rest of the internet”.4 The GPT-3 inter alia writes
poetry, generates tweets, responds to trivia questions, summarizes emails, “translates languages
and even writes its own computer programs”. It can understand the “vagaries of human
language” and is capable of tackling other “human skills”.5 In addition to the above, AI can write
1 See also V.K. Ahuja, “Contemporary Developments in Intellectual Property Rights: A Prologue” in V.K. Ahuja
and Archa Vashishtha, Intellectual Property Rights: Contemporary Development 3-18s (Thomson Reuters, 2020),
2 Rory Cellan-Jones, “Stephen Hawking warns artificial intelligence could end mankind”, BBC News, December 2,
2014, available at: https://www.bbc.com/news/technology-30290540 (last visited on January 23, 2021).
3 Aatif Sulleyman, “Google AI creates its own ‘Child’ AI that’s more Advanced than Systems Built by Humans”,
Independent UK, December 5, 2017, available at: https://www.independent.co.uk/life-style/gadgets-and-
tech/news/google-child-ai-bot-nasnet-automl-machine-learning-artificial-intelligence-a8093201.html (last visited on
January 23, 2021).
4 Cade Metz, “Meet GPT-3. It Has Learned to Code (and Blog and Argue)”, The New York Times, November 24,
2020, available at: https://www.nytimes.com/2020/11/24/science/artificial-intelligence-ai-gpt3.html (last visited on
January 23, 2021).
5 Ibid.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
272
local news articles, generate artwork, write short novels, and generate music by listening to
various recordings.6 AI is also very useful in gaming.
AI has created serious issues and posed challenges in the area of copyright law. The present
article discusses how AI is important in producing creative works such as arts, music, poem,
among other things. The article will also discuss the issues of authorship and deep fakes in the
work produced by the AI.
II. ARTIFICIAL INTELLIGENCE
The term “Artificial Intelligence” was coined by John McCarthy in 1956.7 There is no legal
definition of “artificial intelligence” as of now. “Artificial intelligence” may be stated to be “the
ability of machines to do things that people would say require intelligence”.8 Ray Kurzweil
defined AI in 1990 as “the science of making computers do things that require intelligence when
done by humans”.9 AI normally refers to the “ability of machines to perform cognitive tasks like
thinking, perceiving, learning, problem-solving, and decision-making”. 10 According to Russ
Pearlman, “the central goals of AI include reasoning, knowledge, planning, learning, natural
language processing (e.g., understanding and speaking languages), perception, and the ability to
move and manipulate objects”.11 The three categories of AI systems identified by WIPO are – (i)
“expert (or knowledge-base) systems”; (ii) “perception systems”; and (iii) “natural language
systems”.12
6 Andres Guadamuz, “Artificial Intelligence and Copyright”, WIPO Magazine, October 2017, available at:
https://www.wipo.int/wipo_magazine/en/2017/05/article_0003.html#:~:text=Artificial%20intelligence%20is%20alr
eady%20being,used%20and%20reused%20by%20anyone (last visited on January 23, 2021 ).
7 Fredy Sánchez Merino, “Artificial Intelligence and a New Cornerstone for Authorship”, WIPO-WTO Colloquium
Papers, 2018, p. 28.
8 Philip C. Jackson, Introduction to Artificial Intelligence 1 (Dover Publications, Inc., 1985).
9 Nina Fitzgerald and Eoin Martyn, “An In-depth Analysis of Copyright and the Challenges presented by Artificial
Intelligence”, Ashurst’s Website, March 11, 2020, available at: https://www.ashurst.com/en/news-and-
insights/insights/an-indepth-analysis-of-copyright-and-the-challenges-presented-by-artificial-intelligence/ (last
visited on January 23, 2021 ).
10 Sanjivini Raina, “Artificial Intelligence through the Prism of Intellectual Property Laws” in V.K. Ahuja and Archa
Vashishtha, Intellectual Property Rights: Contemporary Developments 133-41 (Thomson Reuters, 2020).
11 Russ Pearlman, “Recognizing Artificial Intelligence (AI) as Authors and Inventors under U.S. Intellectual
Property Law”, 24 (2) Richmond Journal of Law & Technology 4 (2018).
12 WIPO, “WIPO Worldwide Symposium on the Intellectual Property Aspects of Artificial Intelligence”, WIPO,
March 25, 1991, available at: https://www.wipo.int/edocs/pubdocs/en/wipo_pub_698.pdf. (last visited on January
23, 2021 ).
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
273
The foundation of AI is “artificial neural networks” which are “brain-inspired systems that are
designed to imitate the way the human mind learns”.13 The artificial neural networks have the
capabilities of self-learning that “enable them to produce better results as more data becomes
available”.14 AI therefore, makes a machine to carry out those tasks independently or with
limited human intervention which may otherwise require human intelligence. AI is not to be
considered as one technology, rather a field which has many subfields “such as machine
learning, robotics, language processing and deep learning”.15 “Machine learning” and “deep
learning” are therefore, two subsets of AI.16 For the purposes of machine learning, there happens
to be an inbuilt algorithm in the computer program that “allows it to learn from data input, and to
evolve and make future decisions” either on its own or on the direction. In other words, the
machine learning algorithms learn from the programmer’s provided inputs to generate something
new by making its own independent decisions. Therefore, the parameters are set by the
programmer and the work is generated by the AI itself.17 Most AI examples such as “chess-
playing computers to self-driving cars” can be seen relying heavily on “deep learning”
and “natural language processing”. Computers can be trained with the use of these technologies
to accomplish certain specific tasks including the generation of creative contents by processing
huge data and recognizing certain specific patterns in the data so fed.18
13 Corrs Chambers Westgarth, “Artificial intelligence and copyright: ownership issues in the digital age”, Lexology’s
Website, September 21, 2020, available at: https://www.lexology.com/library/detail.aspx?g=849627a6-c428-4e45-
a386-c6e49d98b446, (last visited on December 1, 2020.).
14 Jake Frankenfield, “Artificial Neural Network (ANN)”, Investopedia, August 28, 2020, available at:
https://www.investopedia.com/terms/a/artificial-neural-networks-
ann.asp#:~:text=An%20artificial%20neural%20network%20(ANN)%20is%20the%20piece%20of%20a,by%20hum
an%20or%20statistical%20standards (last visited on January 23, 2021).
15 Sejal Chandak, “Artificial Intelligence and Policing: A Human Rights Perspective”, 7(1) NLUJ Law Review 46
(2020).
16 WIPO Secretariat, Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence,
WIPO/IP/AI/2/GE/20/1 REV dated May 21, 2020, para 11.
17 Andres Guadamuz, “Artificial Intelligence and Copyright”, WIPO Magazine, October 2017, available at:
https://www.wipo.int/wipo_magazine/en/2017/05/article_0003.html#:~:text=Artificial%20intelligence%20is%20alr
eady%20being,used%20and%20reused%20by%20anyone (last visited on January 23, 2021).
18 Jim Goodnight, “Artificial Intelligence: What it is and Why it Matters”, SAS, available at:
https://www.sas.com/en_in/insights/analytics/what-is-artificial-
intelligence.html#:~:text=Artificial%20intelligence%20(AI)%20makes%20it,learning%20and%
20natural%20language%20processing (last visited on December 3, 2020).
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
274
Two categories of creative works can be produced by the involvement of AI – (i) “AI-generated”
work; and (ii) “AI-assisted” work. The AI-generated works which are also known as “generated
autonomously by AI” refer to the creation of a work by AI without human intervention. In this
category of work, AI may “change its behavior during operation to respond to unanticipated
information or events” and produce work that might not have been intended or anticipated. The
“AI-assisted” works on the other hand are created with significant human intervention.19
III. ARTIFICIAL INTELLIGENCE AND COPYRIGHT
There has been an extensive use of computer programs in the generation of copyrighted works
since the 1970s. The computer generated works did not create much problems with respect to
copyright ownership. The reason was that computer programs were considered mere tools to
support the activities which were creative in nature and the human intervention was required for
the production of the work. These programs were just like stationery items which required
human beings to use them to create works. The things have completely changed now. With AI in
place, the computer programs are no more tools alone and have the potential of generating the
works independently by taking their own decisions.
The AI has the potential to create an enormous amount of work with less investment in a
very short span of time. The works created by AI may qualify for copyright protection in all the
jurisdictions for being original. The requirement of use of “skill and judgement” in originality
may be deemed to have been satisfied by virtue of the “programming and parameter on which
such AI actually compiles and creates the work”.20 However, there will be no author in the case
of AI-generated work. In case of AI-assisted works, there is human intervention. Therefore, in
case of latter, the person who caused the work to be created by using artificial intelligence may
claim himself to be the author, but the same is not true where the work has been created by AI
itself without any human intervention. The issue of authorship in such cases has puzzled all
countries of the world. There can be three broad possibilities with respect to the authorship issue
19 WIPO Secretariat, Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence,
WIPO/IP/AI/2/GE/20/1 REV dated May 21, 2020, para 12.
20 Lucy Rana and Meril Mathew Joy, “India: Artificial Intelligence And Copyright – The Authorship”, Mondaq,
December 18, 2019, available at: https://www.mondaq.com/india/copyright/876800/artificial-intelligence-and-
copyright-the-authorship (last visited on December 1, 2020).
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
275
– (i) the copyright system should recognize authorship for AI; (ii) there should be no authorship
in AI-generated work and the work should fall into the “public domain”; and (iii) there should be
sui generis law rather than copyright law to protect such works.
The copyright protection serves as an incentive for the author to produce more creative
works using his skills, labour and judgement. If the AI is recognised as an author and the AI-
generated works are protected under the copyright law, then it would mean that “human
creativity” and “machine creativity” are on the same pedestal. On the other hand, if AI-generated
works are not protected by copyright law, then it would necessarily mean that human creativity is
preferred over machine creativity. Preferring machine creativity over human creativity or putting
both at the same pedestal is likely to kill human creativity in the long run.
Considering AI as an author of the AI-generated work may cause several issues. The
work generated by AI may not be flawless. The AI may use biased and toxic language21 which
may result in defamation or obscenity; incite violence on the lines of caste, creed or religion; or
produce any other undesired result. In such a scenario, it will be difficult to fix the civil and
criminal liability of the AI as it has not been recognized as a person. At the most such work may
be deleted or in worst cases that AI software may be banned, but till then, it may be too late and
irreparable damage might have been caused by that work. Another issue is that if the AI-
generated work happens to be “substantially similar” to an existing work which may have
copyright, how will the AI be held as an infringer in such a case? Further, if AI is treated as an
author, it will not be entitled to transfer ownership in the work, in absence of personhood.
The premise which reflects from civil law countries such as Germany, France and Spain
indicates that works created must bear the “imprint of the author’s personality”. The authorship
therefore, should be denied to AI in the AI-generated works as the AI does not have
personality.22 Making AI a legal entity would mean that it should possess the capacity to enter
into contracts with other persons. It will also have duties under the law and will be liable for its
21 Supra note 4
22 Brigitte Vézina and Brent Moran, “Artificial Intelligence and Creativity: Why We’re against Copyright Protection
for AI-Generated Output”, Creative Commons, August 10, 2020, available at:
https://creativecommons.org/2020/08/10/no-copyright-protection-for-ai-generated-output/ (last visited on January
23, 2021).
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
276
acts. Most importantly, it should have the capacity “to sue and be sued” under the law. Most of
the countries are not in a favour of granting legal status to AI.
It will however, not be out of place to mention that the European Parliament has
advocated to grant the legal status of “electronic persons” to “autonomous robots” for the
purposes of protection under copyright law.23 It may also be added that Artificial Intelligence
Virtual Artist (AIVA) Technologies’ “music composing AI becomes the first in the world to be
officially be given the status of a composer”. It has been recognised officially as a composer
by “SACEM, France and Luxembourg author’s right society” which enables it to release music
and get royalties under the name AIVA.24 It is also noteworthy that Saudi Arabia has granted
citizenship to an AI humanoid robot, Sophia in 2017. Dr. David Hanson, who created Sophia
writes in his paper titled “Entering the Age of Living Intelligence Systems and Android Society”
that looking at the progress in AI, there will come a tipping point where robots will awaken and
insist on their rights to exist, to live free, and to evolve to their full potential”. This also means
that they will be claiming intellectual property protection with respect to Intellectual Property
Rights (hereinafter as “IPRs”) they will create. According to him, “advanced robots will have the
right to marry, own land and vote in general elections by 2045”.25
It is noteworthy that the copyright laws of many countries also provide moral rights to the
author, though this is not an obligation under the TRIPs Agreement.26 Two moral rights – (i)
right of paternity; and (ii) right of integrity are ordinarily provided to the author. The former
ensures the right of the author to be associated with his/her work and be named as its creator,
whereas the latter enables the author to claim damages for any mutilation or distortion of the
work if that is prejudicial to his/her honour or reputation. In Amar Nath Sehgal v. Union of
23 Edward Klaris and Alexia Bedat, “Copyright Laws and Artificial Intelligence” , American Bar Association,
November 16, 2017, available at:
https://www.americanbar.org/news/abanews/publications/youraba/2017/december-2017/copyright-laws-and-
artificial-intelligence/ (last visited on January 23, 2021).
24 Ed Lauder, “Aiva is the first AI to officially be Recognised as a Composer”, AI Business, October 3, 2017,
available at: https://aibusiness.com/document.asp?doc_id=760181 (last visited on January 23, 2021).
25 Anthony Cuthbertson, “Robots will have Civil Rights by 2045, Claims Creator of ‘I will Destroy Humans’
Android”, Idependent UK, May 25, 2018, available at: https://www.independent.co.uk/life-style/gadgets-and-
tech/news/robots-civil-rights-android-artificial-intelligence-2045-destroy-humans-sophia-singularity-a8367331.html
(last visited on December 3, 2020).
26 Trade Related Aspect of Intellectual Property Rights, art. 9.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
277
India,27 the Delhi High Court observed that “in the material world, laws are geared to protect the
right to equitable remuneration. But life is beyond the material. It is temporal as well. Many of us
believe in the soul. Moral rights of the author are the soul of his works. The author has a right to
preserve, protect and nurture his creations through his moral rights”. Moral rights are related to
the feelings and emotions of the human author. These rights are not meant for AI.
Another perplex question will be about the term of AI-generated works. The AI does not
die like a human being. One may however, argue that the term may be counted from the date of
publication for a period of 50 or 60 years depending upon the laws of the countries. Conferring
copyright protection on AI with respect to AI-generated works is disputed on the ground that a
human being is mortal and experiences fatigue while working. Therefore, a human author creates
limited works in his/her lifetime in which a copyright subsists, and the copyright is justified as
his/her efforts are to be rewarded. On the contrary, an AI is immortal, does not experience
fatigue and is capable of producing any number of works. Conferring copyright protection on
AI-generated works is therefore “equivocal and disputable”.28 Further, the experts who are not in
favour of giving copyright protection to AI-generated works argue that if the same model and
same inputs are given, the AI will produce output that will be the same every time. Therefore, it
is hard to say that it is “unique and creative”.29
In addition, it will be difficult for AI to negotiate the royalty with others and enforce the
rights which are available to the author under the copyright law. Making AI an author of the
work will not be an easy task as it is likely to create more difficulties than to resolve them.
Another view that emerges from the discussion is that there should be no authorship in
AI-generated work and the work so created should fall into the “public domain”. There are
several justifications for putting the AI-generated works in public the domain. One of the reasons
is that while producing a work by AI, no cost is incurred by it, therefore it is quite logical to
make the AI-generated work accessible to the public for free. Secondly, AI is capable of creating
27 2005 (30) PTC 253 (Del).
28 Sik Cheng Peng, “Artificial Intelligence and Copyright: The Author’s Conundrum”, WIPO-WTO Colloquium
Papers,181 ( 2018).
29 Supra note 22.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
278
any number of iterations of work created by it for no extra cost or resources. Last but not the
least, one of the objectives of copyright law is to provide an incentive to the author of the work
in terms of providing economic rights and moral rights to motivate him to produce more works
for the advancement of society. The AI, being non-human does not require any such motivation
to create the work.30
One should however, also consider the fact that if there is no protection to AI-generated
works and the public is free to make use of such work without any authorization or paying any
fee, it may turn out to be a death knell for those companies which invest a huge amount in the AI
system to generate these works. Smart people will start commercializing such works in various
ways without incurring any cost and compete with companies which invested the money.
Therefore, some protection may be needed for AI-generated works to encourage the AI
programmers and the companies which may work as a stimulus for them to continue investing in
the AI related R&D activities.31
The UK Copyright, Designs and Patents Act, 1988 (hereinafter referred to as “CDPA”)
deals with computer-generated work. “Computer-generated” work is defined under CDPA to
mean that “the work is generated by a computer in circumstances such that there is no human
author of the work”. 32 The reason for such a provision is “to create an exception to the
requirement of human authorship in order to provide due recognition and protection for the work
that goes into creating a program capable of independently generating works”.33 According to
section 9(3) of the CDPA, the author in the case of a “literary, dramatic, musical or artistic work
which is computer-generated” is to be “taken to be the person by whom the arrangements
necessary for the creation of the work are undertaken”.
30 Ayush Pokhriyal and Vasu Gupta, “Artificial Intelligence Generated Works under Copyright Law”, 6(2) NLUJ
Law Review 116 (2020).
31 Pamela Samuelson, “Allocating Ownership Rights in Computer-Generated Works” 47 University of Pittsburgh
Law Review 1185 (1986).
32 The Copyright, Designs and Patents Act, 1988, s.178.
33 Supra note 9.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
279
Andres Guadamuz states that authorship in such a case goes to the programmer and not to
the user. He explains this position by giving the example of Microsoft which developed the
computer program “Word” for the users to create their own works. Microsoft cannot have a
copyright in a work which has been produced by the user with the help of that program. The
copyright in the work so created will lie with the user of the program who will be recognized as
the author because he/she created the work using that program.34 In Express Newspapers plc v.
Liverpool Daily Post & Echo,35 the court considered computer as a tool in the same manner as a
pen is regarded as a tool. In the United States also, the author of a work which is created with the
help of AI may have copyright if he/she establishes that the AI program was used as a
tool/medium in the creation of the work.36 In Naruto v. Slater,37 popularly known as “Monkey
Selfie” case, the court in the United States held that the monkey could not be taken as the author
of the selfies it clicked. Copyright in a work can only be conferred on a human author and not on
animals and machines in the U.S.38
The situation however, shall be different in the case of “artificial intelligence algorithms”
which have the potential of generating work on its own. Where computer using AI acts as
an “independent actor” and generates works “algorithmically, sequentially, or non-
deterministically”, there appears an “apparent gap between the human’s input and the computer’s
output”.39 The contribution of the user in generating the work in such a scenario may not be more
than pressing a button which enables the machine to create that work. Therefore, in such a case,
“the person making the arrangements for the work to be generated” should be taken as a
programmer.40 It can also be argued in the manner that “an assumption can be derived that the
programming of the AI is made in such a manner that it can create and identify equations to
generate a result on its own, and therefore, the creativity may vest with the programmer who has
created the AI, with sufficient programming”.41
34 Supra note 17.
35 [1985] FSR 306.
36 Kalin Hristov, “Artificial Intelligence and the Copyright Dilemma”, 57(3) IDEA 435 (2017).
37 2016 U.S. Dist. Lexis 11041 (N. D. Cal. Jan. 23, 2016).
38 Supra note 36 at 449.
39 Nahide Basri, “The Question of Authorship in Computer-Generated Work”, Penn Law, University of
Pennsylvania, January 13, 2020, available at: https://www.law.upenn.edu/live/news/9691-the-question-of-
authorship-in-computer-generated. (last visited on January 23, 2021).
40 Supra note 17.
41 Supra note 20.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
280
Section 9(3) of the CDPA may be given a different interpretation as suggested by Sik
Cheng Peng in his study. He argues that where a user takes part in the selection of data that is to
be fed to the AI system, then the user should be considered the person who initiated the process
to create the work. The user, therefore should be taken as the person who made the “necessary
arrangements” to create the work and not the AI or the programmer or the company owning the
AI. Consequently, the user should be assumed to be the author of the AI-generated work as
opposed to AI or the programmer.42
The Indian Copyright Act does not define “computer-generated work” like CDPA. It
however defines “author” in relation to “any literary, dramatic, musical or artistic work which is
computer-generated” as “the person who causes the work to be created”.43 In Camlin Pvt. Ltd. v.
National Pencil Industries,44 the Delhi High Court elaborated the meaning of the term “author”.
The Courts stated that “mechanically reproduced printed carton” was not a subject matter of
copyright for the reason that it was not possible to determine who the author of such carton was.
The Court further stated that “copyright is conferred only upon authors or those who are natural
person from whom the work has originated. In the circumstances the plaintiff cannot claim any
copyright in any carton that has been mechanically reproduced by a printing process as the work
cannot be said to have originated from the author. A machine cannot be an author of an artistic
work, nor can it have a copyright therein”.45 In Tech Plus Media Private Ltd v. Jyoti Janda,46 the
Delhi Court held that “the plaintiff is a juristic person and is incapable of being the author of any
work in which copyright may exist”. The Court further stated that the plaintiff, however, could
become the owner of the copyright in the work under a contract with its author.47
In Australia, copyright is available to the creator of an AI machine in the “machine’s
source code” only but not in the AI-generated work because of the lack of human intervention.48
42 Supra note 28.
43 The Copyright Act, 1957, s.2 (d) (vi).
44 AIR 1986 Delhi 444.
45 Id., at para. 54-55.
46 Delhi High Court, September 29, 2014; available at: https://indiankanoon.org/doc/18739837/ (last visited on
January 23, 2021).
47 Id., at para 20.
48 Supra note 9.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
281
The best approach is to decide the issue of authorship in a case on its merits by adopting a case-
by-case approach.
The suggestion of considering AI and human author as joint authors of the work so
produced is not a sound one. The reason is that human beings do not control all operations of AI
and the AI operates without any control. This does not fit into the definition of “works of joint
authorship”. For example, the Indian Copyright Act, 1957 defines “work of joint authorship” to
mean “a work produced by the collaboration of two or more authors in which the contribution of
one author is not distinct from the contribution of the other author or authors”.49 Rich points out
that “machine learning tends to create models that are so complex that. Even the original
programmers of the algorithm have little idea exactly how or why the generated model creates
accurate predictions.’50 In addition, the suggestion of making AI programmer and the AI user
joint author of AI-generated program is also not a sound one.
At the international level, the Berne Convention, 1886 did not contemplate “non-human
authorship”.51 The same position may be considered as correct in the case of Trade Related
Aspects of Intellectual Property Rights (hereinafter as “TRIPs”) Agreement in view of the fact
that it incorporates the provisions of the Berne Convention. A similar position may be deemed to
be correct about the WIPO Copyright Treaty and WIPO Performances and Phonograms Treaty
of 1996 (WIPO Internet Treaties). At the same time, it may also be argued that the international
legal regime on copyright did not preclude the possibility of a non-human authorship in national
legislations.52 The international treaties ordinarily lay down minimum common standards to be
followed. The countries are obligated not to derogate from them, but at the same time, they are
free to provide better protection than what has been laid down in the treaties.
49 The Copyright Act, 1957, s.2(z).
50 Michael Rich, “Machine Learning, Automated Suspicion Algorithms, and the Fourth Amendment” 164 University
of Philadelphia Law Review 871, 886 (2016).
51 Sam Ricketson, “People or Machines: The Berne Convention and the Changing Concept of Authorship”, 16(1)
Columbia VLA Journal of Law and the Arts 1 (1991)
52 Dilan Thampapillai, “The Gatekeeper Doctrines: Originality and Authorship in Australia in the Age of Artificial
Intelligence”, WIPO-WTO Colloquium Papers, 2019, p.no. 2.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
282
The AI-generated works may also be protected outside the copyright system by adopting
a sui generis system. Such a system may have lesser protection in terms of copyright duration
and other things. One author has suggested that duration of such works may be set as low as 5 to
10 years. He argues that within the copyright system, by providing protection for shorter duration
“the new model of AI copyright protection would give rise to significantly less interference with
the existing norms of copyright law. There would be lesser potential for AI authors to crowd out
human authors in creative markets, as the former would soon lose their copyrights”.53 Sik Cheng
Peng suggests that if AI-generated works are to be protected, the same may be protected under a
sui generis right similar to one conferred on “databases” under the European Union Database
Directive.54 The right so granted may prevent “outright and unfair exploitation of the works”.55
Such a system may have drawbacks such as non-disclosure of involvement of AI by its
owner. The system, therefore must ensure that there is true disclosure with respect to the creation
process of such works and the involvement of AI. The law relating to unfair competition may
also be an option for the protection of the AI-generated works.56 The World Intellectual Property
Organizations (WIPO) is already considering the issue of authorship and the law which may be
adopted in this regard.
Apart from the issue of authorship, a new issue has been identified by WIPO, i.e.
copyright issue in “deep fakes”. “Deep fakes” is basically “the generation of simulated likenesses
of persons and their attributes, such as voice and appearance”. The role of AI in the deep fakes
technology is ever increasing. There may be more issues than copyright alone, such as privacy,
defamation, etc. when somebody is shown in deep fake without his/her authorization and the
actions and views of the person shown in the audio-visual work are not authentic. The deep fakes
audio-visuals of the popular players, performers, leaders, and other known personalities may
become very popular among the public and may have a very good market. These deep fake
works may also continue after the death of such persons and bring good revenue to its creators.
53 Id., at 6.
54 Directive 96/9/EC, of the European Parliament and of the Council of March 11, 1996 on the Legal Protection of
Databases, 1996 O.J. (L 77) 20.
55 Supra note 28/
56 Supra note 22.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
283
The moot question that arises here is whether such deep fake work, if produced without
the authorization of the person concerned should be protected under the copyright law at all.
Further, where authorization has been given by the person concerned, what will be his/her rights
in such works under the copyright law? Can a system of equitable remuneration for the creator of
deep fakes and the persons concerned depicted in the work be adopted?57 These issues have to be
resolved as increasing uses of AI will continue to pose more challenges in the time to come. The
efforts are also being made to resolve the aforesaid issues at the WIPO.58
IV. ARTIFICIAL INTELLIGENCE AND DATA PROTECTION
Data are an extremely important component of AI applications. The reason is that such
applications rely upon “machine learning techniques that use data for training and validation”. If
more data is available for machine learning, the results are likely to be better, authentic and
refined.
The creative works may be produced by an AI application that learns from data that has
been used to train such an AI application. The data so used may be economically valuable and
may have copyright protection. The important question which arises is whether the use of such
data for the purpose of machine learning without authorisation from the copyright owner
amounts to copyright infringement. If yes, then how to enforce such copyright? Further, can a
general exception be made in the copyright law with respect to the use of data for the purpose of
machine learning? or alternatively, whether such exception should be confined to “non-
commercial user-generated works” or for the purpose of “research”? Another mind-boggling
question which may arise is that if an AI application automatically produces a work which is
similar to the original work contained in the data used for machine learning, would this amount
to an infringement of copyright? If yes, who is going to be the infringer in such a case and how
the copyright will be enforced? On the contrary, should there be “free flow of data” so that
improvisation could be made in the AI?59
57 WIPO Secretariat, Revised Issues Paper on Intellectual Property Policy and Artificial Intelligence,
WIPO/IP/AI/2/GE/20/1 REV , May 21, 2020,
58 Id., at para 23.
59 Id., at para 24.
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
284
In response to the aforesaid questions, it will be appropriate to refer to the doctrine of fair
use/dealing. Where the economic value of the copyrighted work used for machine learning has
been reduced to its owner due to AI created work, it may not be considered as fair use/dealing. If
it does not result in the reduction of the economic value of such work, then it may qualify as fair
use/dealing depending upon the national laws of countries. Ordinarily, the economic value of
copyrighted material used in training algorithms is not affected. Therefore, if a work is created
using an algorithm powered tool, which is completely different from the copyrighted material
used for machine learning, the economic value of the latter is unlikely to change.60
On the analogy of Google Book case, it may be suggested that the use of “copyrighted
works for the non-expressive purpose of training AI models amounts to fair use”. It is
noteworthy that Japan has amended its copyright laws and included “exemptions of the use of
copyrighted works for machine learning”.61
It is also important to note that the “selection or arrangement of data”, being an
intellectual creation may be a subject matter of protection under copyright or a sui generis law
under various jurisdictions. The data contained in such compilation may or may not be
copyrighted. Looking at the important and increasing role played by AI, it is sine qua non to
have a legal framework on data protection from the point of view of ascertaining the authorship
in case of creative works and inventorship in case of inventions. Such a law is also needed for
promoting creativity and innovation and for assuring fair market competition in the society. The
law needs to have a balanced approach because the over protection of data may adversely affect
the machine creativity which is likely to dominate the field of creativity in the future. India,
unfortunately, does not have a data protection law as of now. However, “computer programmes,
tables and compilations including computer databases” are protected in India under Copyright
Act, 1957 as “literary works”.62
60 Karen Robinson, “Copyrights in the Era of AI”, Adobe Blog, February 27, 2020, available at:
https://blog.adobe.com/en/publish/2020/02/27/copyrights-in-the-era-of-ai.html#gs.opdukw (last visited on January
23, 2021).
61 Ibid.
62 The Copyright Act of India 1957, s.2(o).
Electronic copy available at: https://ssrn.com/abstract=3905190
ILI Law Review Winter Issue 2020
285
V. CONCLUSION
The role of AI is going to increase in all sectors in our day to day lives by leap and
bounds. The law is required to regulate its uses. In case of intellectual property rights,
particularly in copyright, AI will continue to play an extremely important role. The issues of
authorship and ownership of AI-generated works in copyright law have forced the international
community to think and devise an acceptable solution for all countries. There is no foolproof rule
to address this issue and every rule has its own flaws. There will be significant ramifications for
offering non-human authorship to AI-generated works. Putting the AI-generated works in the
public domain is also not a good idea as it will discourage the AI programmers and companies
owning such AI to further invest in the AI domain. The WIPO is working hard to address these
issues. The sui generis system may be a better option or alternatively, some provisions in the
copyright legislations of the countries which are specifically drafted for AI and AI-generated
works may address this issue. In any case, the AI-generated works should be provided lesser
protection and human creativity should be preferred over machine creativity. A balanced
approach is therefore, the need of the hour.
Electronic copy available at: https://ssrn.com/abstract=3905190