Creative robots, powered by powerful artificial intelligence systems and
working autonomously, can develop novel artistic and other works that would be
copyright protected if made by humans. This article discusses innovative robot
artwork copyrightability. We believe autonomous creative robots can own
copyrights to their creations. Copyright law's most essential principle,
originality, poses the biggest obstacle to this idea.
This article revisits uniqueness. This article contends that the new 3A
era�advanced, automated, and autonomous�requires a re-evaluation of originality,
which is undefined under copyright protection under the Berne Convention. The
absence of a precise definition makes the idea of originality unsuitable for
creative robots' potentially "copyrightable" works.
The business and public are already confused by ambiguous, immeasurable, and
disharmonized definitions of "originality." An key legal notion is unknown,
prompting a quest for a remedy to minimise or lessen future problems. This
article proposes a formal, objective approach to originality instead than the
current subjective (or mixed) one. The nebulous, imprecise character of art
supports the proposed objective approach to creative robot and digital tool
artworks. A formal, objective, descriptive method is needed to answer the legal
issue "what is original?"
The use of artificial intelligence (AI) in the production of any particular
piece of artistic work has always been a topic of discussion among those
necessary parties. Generative AI systems and the artwork they produce raise
important questions that border on authorship, licencing, and reproductive
rights, amongst others. These questions stem from general policy concerns as
well as the individual interests of the artist who was involved in the creation
of the work. The use of AI to create artworks also creates some contentious
issues because it could potentially infringe on the intellectual property rights
If, for example, a creative work is based on machine learning, a technique that
enables an AI to learn for itself from data that it has been provided, then this
perspective can be considered genuine. Shouldn't the person who produced the
input (in this case, the data) on which a production is based also be the owner
of the work that is produced as a result of that input? And even if the work
were to be generated by an AI on its own, could it really be considered to be
Copyright laws have traditionally been based on a human creator who creatively,
originally, and independently creates a work of art from the depth of his or her
imaginations in such a way that it reflects his or her personality, ideas, and
belief systems. This has been the case from the perspective of copyright laws
across multiple jurisdictions.
This article seeks to critically examine the legal issues involved in the
copyrightability of artificial intelligence art using a multi-jurisdictional
approach, citing unique challenges the situation creates, and suggesting
possible areas of legal reform in order to deal with this eventuality that is
quickly becoming pervasive in our world today. In order to reconcile this
divergence, this article seeks to critically examine the legal issues involved
in the copyrightability of artificial intelligence art using a
AI Or Computer-Generated Artworks
Artists have used computers to make digital art for 50 years, but they had to
write code to achieve the ultimate aesthetic work. In recent years, artists have
used computers to learn this aesthetics and apply them to hundreds of images.
Generative Adversarial networks are used in most AI artworks. Ian Goodfellow's
2014 discovery allowed researchers to make realistic computer-generated pictures
of people's faces. Although this technology has been infrequently used to
make contentious "deepfake" movies, it has led to more advanced AI that can
accurately recreate photos and drawings.
AI favours spectacle and kitsch, and a few artists and industry professionals
believe machines cannot create like humans. "Replicating artistic skills
involves also simulating human thinking and reasoning, especially creative
thinking," say opponents of computational creativity. Algorithms and information
processing systems cannot do this. They say this is because these algorithms
can't capture the human mind's brilliance and creativity, and these works are
developed with a lot of human input. An AI artwork sold at Christie's Auction
House in New York for $432,500 was 45 times its anticipated price of
The Next Rembrandt is another AI art piece that has sparked scholarly debate on
the copyrightability of AI-generated works by human artists. The Next Rembrandt
was created by data scientists, engineers, and art historians who used
traditional data analysis to deeply analyse and recreate Rembrandt's painting
styles, style, and subject matter and transfer that knowledge into software that
could generate the new work using the latest 3D printing technology.
AI systems are more autonomous than ordinary software. Complex AI systems can
accomplish tasks without human input. In order for us to understand this
feature, we must emphasize some familiar tools these systems require to produce
these works. A paintbrush, even if it is helpful in creating an artwork has very
little to do with the end result (the artwork) and the user of the paintbrush is
the owner of the artwork. Diverse tools are needed to make different goods, but
the user (author) is still accountable for the creative contribution.
Machine learning also helps AI systems replicate familiar art styles. Machine
learning is a sub field of AI that may be defined as the study of algorithms and
systems that improve their performance on a particular job when they are
provided with additional data. The performance of AI systems based on machine
learning relies crucially on the quality of the data. Thus, the data provider
must be acknowledged.
Understanding an AI system's autonomy, especially regarding copyright, is
crucial Hard work. AI systems have various degrees of automation, depending on
how much control the human operator has. To identify such levels of
independence, "in the loop," "on the loop," and "out of the loop" are commonly
used interchangeably. An art-creating robot may need a human operator to learn
from prior paintings and be monitored during implementation . Technological,
social, legal, and ethical variables may influence the correct degree of
The Concept Of Authorship In Ai Or Computer-Generated Art
AI-created art appears to present pertinent questions about the copyrightability
of some art genres. This is strongly related to the fact that, in some
jurisdictions, copyright ownership can only be vested in a human author under
particular copyright legislation. This reasoning appears to be based on the
highly recognized US case of Naruto v. David John Slater, sometimes known as
the "Monkey Selfie case." Naruto, a crested macaque, snatched up a
photographer's camera and took photos of himself.
The photographer, David Slater, and Wildlife Personalities Ltd released the
"Monkey Selfies" pictures in a book and claimed copyright ownership of the
images. People for the Ethical Treatment of Animals (PETA) and Dr Antje
Engelhardt sued on Naruto's behalf as next friends, saying Naruto was the
creator of the images and that Slater and Wildlife Personalities violated
Naruto's copyright. Naruto's claim was rejected, and the Ninth Circuit upheld,
ruling that Naruto lacked standing to sue under the Copyright Act since animals
cannot sue for infringement.
The court reasoned that "Section 101 of the Copyright Act does not clearly
empower animals to bring copyright infringement claims under the legislation."
Although corporations can hold copyrights and sue under the Copyrights Act, the
court emphasized that under US Supreme Court law, companies are considered
"persons," and unlike animals, these organization's "are founded and owned by
Because of an earlier Supreme Court rule that copyright law exclusively protects
"the products of intellectual effort and are built on the creative capacities of
the human mind," the US Copyright Office will not approve a registration unless
the author is a human person. According to the Copyright office, this excludes
"works created by a machine or mere mechanical process that functions randomly
or mechanically without any creative input or interaction from a human
Human authorship was also emphasized as the standard for copyright protection in
the Australian case of Telstra Corporation Ltd v. Phone Directory Property
Ltd, where humans were involved in gathering and organizing the collection
of materials for phone directories, but this collection phase was not geared
toward the creation of the work. The extraction step of sorting the directories
in alphabetical order and transferring the material into electronic form was the
Because there was no human participation in the extraction step, it was
determined that copyright did not exist in the telephone directories. The Full
Federal Court emphasized in this instance the necessity that "the work must
begin with an author or joint writers." The lack of a human author behind
Telstra's directory collection resulted in the absence of copyright protection
in the directories.
European copyright law, which consists of international copyright treaties to
which the EU is a party, EU legislation, and cases at the Court of Justice of
the European Union (CJEU), appears to have followed this general trend, though
EU national governments have discretion to import their respective
interpretations on authorship of creative works. Notable international treaties
include the Berne Convention, the World Intellectual Property Organization's
Copyright Treaty, and the TRIPS Agreement.
The degree of intellectual production necessary to achieve the requirements of
originality was a major topic considered at the commencement of the Berne
Convention. The degree of expertise and work involved was often stressed in
common law systems, whereas the amount of inventiveness was valued as a basis in
civil law systems.
The Berne Convention, on the other hand, defines an author by stating that if
the author's name is included, he or she is presumed to be the creator of a
literary or creative work in the absence of proof to the contrary. As a result,
it is clear that the purpose of this regulation is to provide some certainty
while also lessening the burden of evidence for right holders.
The WIPO Copyright Treaty and the TRIPS Agreement, on the other hand, are
quiet on the concept of 'author,' despite the fact that both accords demand
conformity with the Berne Convention. Furthermore, the European Union's Court of
Justice has stated on numerous occasions, most notably in its landmark decision
of Infopaq International A/S v. Danske Dagbaldes Forening
copyright only applies to original works, and that originality must reflect the
"author's own intellectual creation." This is interpreted to signify that an
original work must represent the creator's personality, implying that a human
author is required for a copyright work to exist.
However, there has been an increasing push, particularly in Europe, to shift the
conventional perception of human authorship since the Naruto case. This new
approach is best exemplified by Section 9(3) of the Copyright, Designs and
Patents Act 1988 (CDPA), which states that "in the case of a literary, dramatic,
musical, or artistic work that is computer-generated, the author shall be taken
to be the person by whom the arrangements necessary for the creation of the work
are undertaken". Furthermore, section 178 of the CDPA defines a computer
produced work as one that "is generated by computer in conditions such that the
work has no human author".
The reasoning for this rule is to make an exemption to all human authorship
criteria, so honouring the industry that goes into developing a programme
capable of generating art forms, even if that creative impulse is driven by a
computer. These views were echoed in the Irish Copyrights and Related Rights Act
No 28/2000 (S.2(1)) and the New Zealand Copyright Act 1994 as modified (S.2).
The upshot of this important revision was to broaden the definition of human
authorship to include those who created the arrangements for a robot or machine
to create any specific work. This pertains to the previously described human in
the loop scenarios.
Can Creative Robots Be The Authors Of Artworks?
Creative robots already produce a tremendous number of creative works which
could be copyrightable. Imagine a portrait produced by a creative robot. The
artist is not a human being, but rather a robot driven by artificial
intelligence systems. The subject of the portrait is unknown to the human
operators of the AI system until the creative robot has generated the image.
Such a creation cannot meet the subjective concept of originality. However, the
portrait might meet the concept of objective originality suggested by this
article and hence, would be copyrightable.
Subject Matter Eligibility
As one of us explains in another piece, creative robots are autonomous,
creative, independent, and unexpected, making them qualified for authorship.
Copyright protection requires a minimum of imagination, focus, and, most
importantly, originality. The work must be at least creative. The job must be
tangible. Third, it must be vaguely original. A work must meet all three
criteria to be copyrightable. Only copyrightable works fulfil all three
Creative robots can get copyright protection if these requirements are met.
Inventions by AI-controlled robots are Copyrightable. Creative robots readily
match the first two requirements, but originality may be difficult, especially
given the subjective nature of uniqueness. Eligible subject matter should
receive intellectual property rights. 44 Because of their basic 10 traits and
the objective definition of uniqueness, creative robots may easily meet all
three requirements. Creative robots may get copyrights on their ideas if they
can generate new works without intention.
THE MISUNDERSTANDING OF THE CURRENT (TRADITIONAL) COPYRIGHT REGIME
Copyright law presently excludes creative robots' works. We argue that the old
legal system exclusively considers human authors behind the creative process.
Policymakers have evaluated the human behind the creative robot's works and
pronounced that person the creator and entitled to copyright protection when
using this traditional method.The standard method misunderstands creative
robots' autonomous, independent creation of new works. Creative robots require
rethinking "originality." The objective approach to originality is important to
identify creative robots as legal entities, like corporations, which can possess
The Programmers Of AI Software And The Artworks Produced By Creative Robots
The AI software programmers might argue that creative robots' creations are
theirs. Following the old method of identifying the human author behind the
innovative robot, the programmer or operator would receive copyrights.
Traditional artists have copyright immediately. Many contributors create the
algorithms that fuel the next generation of creative AI systems. However, the
creative robot may create its own works. If humans designed the AI software
programme, may the programmer claim copyright on AI system-generated works
Are Artworks Produced By Creative Robot's Derivative Works?
Rearden LLC and Rearden Mova LLC's case against The Walt Disney Company posed
the question of whether creative advanced technology system-produced art is
derivative. The complaint argues that Rearden, which owns Mova technology, owns
derivative works like the Beast's animal-like visage from Beauty and the Beast,
which was created utilising unique Oscar-winning effects employing MOVA.
"Disney used the stolen MOVA Contour systems and methods, made derivative works,
and reproduced, distributed, performed, and displayed at least Guardians of the
Galaxy, Avengers: Age of Ultron, and Beauty and the Beast, in knowing or
willfully blind violation of Rearden Mova LLC's intellectual property rights,"
the lawsuit states. The derivative works paradigm assumes that a computer
programme or system's output is a derivative work of the proprietor of the
copyrighted programme or patented system. The copyright holder also owns the
output under U.S. law.
This conclusion on creative robots' production assumes that computers need a lot
of human input to create art. Congress created the National Commission on New
Technological Uses of Copyrighted Works (CONTU) to address computer-generated
work creation. CONTU determined that the user should have authorship rights if
they significantly contribute to the computer's output. However, complex,
advanced AI systems offer more intricate and challenging challenges than
traditional copyright law.
This is because creative robots' creations are owned separately from the robots
and technology. When the computer creates most of the output independently and
creatively without duplicating or violating other writers' rights, the owner is
less likely to be deemed the original source. Thus, Rearden's derivative works
technique may fail since we don't know if the works are derivatives or fresh and
original. If a human provided input that altered the artwork, such as by
introducing the robot to fresh data, the finished piece may be the result of
another human's creative involvement.
Thus, this human, not the programmer or robot owner, may have copyright
protection for the composition. Disney's cinematographer, director, and actors
created the Beast and others. We don't dispute the programmer's copyright
claim in this piece. However, we disagree that the programmer owns the AI
system's products and processes. Can robots create art? Originality might
determine the case.
Copyright legislation can be nullified as useless and outmoded. However, if we
keep the copyright law, we should reassess threshold notions like originality
and give them new definitions to accommodate the 3A (advanced, automated, and
autonomous) age and, in our debate, a digital author of artworks as the
original. This approach may better reflect creative robots' behaviour. First and
first, we must redefine originality.
Licensing And Other Related Rights In Computer Generated Or AI Art
The development of AI systems being able to efficiently duplicate works of art
with exceptional precision utilizing data acquired from earlier works has raised
concerns about the import of licencing and its impact on the completed piece.
The Next Rembrandt was a case in point, in which machine learning algorithms
assessed old artwork by the Dutch great artist and developed a new portrait in
the Rembrandt style.
The licencing issue appears to have been resolved due to the fact that existing
copyright on the original work may have elapsed due to time and the industry
used by the AI programme to generate a duplicate, so making it an original
artwork in its own right.
Scholars claim that the work of AI systems is highly unexpected because the
system continually and automatically evolves with additional data. In short, AI
systems are not only more precise, of greater quality, and faster at processing
information due of their intelligence components, but they are also capable of
producing unanticipated, unique, and creative artworks and other goods that are
unknown to their programmers. In a nutshell, they may have the effect of
significantly modifying the completed work so that it bears little or no
resemblance to the original.
However, AI artists that build on prior art face the dual challenges of
appropriation and inspiration, and where a line must be set, particularly when a
computer programme is given reproductions of earlier artworks. However,
prior appropriation rulings in the United States indicate that some uses may be
protected under fair use provided the outcome is sufficiently transformative
from the source work. As a result, if an AI artist sells or exhibits AI art that
is significantly similar to the underlying work, the AI artist is unlikely to be
able to rely on fair use.
As a result, AI artists who want to sidestep the muddy waters of fair use and
transformative rights should only employ public domain materials. AI artists
might also consider analyzing their creations to see if they are sufficiently
transformative before releasing them to the public to avoid any infringement
claims. However, I believe that if an art piece is sufficiently transformative
that the completed product looks nothing like the source, it should be granted
copyright protection not for the programme itself, but for the person(s) who
produced the programme.
AI art licencing has come a long way, and with new template licence and
collaboration Agreements for AI art recently launched by the Harvard Law
School's Cyber Law Clinic, efforts are now being made to protect these artworks
from possible infringement and educate AI artists on the importance of
protecting their works.
Indian Perspective For Protection Of Computer-Generated Works
AI generated art is in full swing and the Indian Copyright Office is confused.
AI generated art is created autonomously by artificial intelligence without
creative contribution from humans . Such works qualify as 'computer generated
works' under the Indian Copyright Act. Computer-generated works were included as
a category of works in 1995, presumably at a time when AI was not making art.
For such works, copyright law confers authorship to the "person who causes the
work to be created". The question of whether AI can be named as an author of a
work is debatable. There is no Indian precedent to suggest that authorship can
be conferred to computer / software. There is also no policy guidance from the
Indian Copyright Office on whether only humans can be considered authors.
The Indian Copyright Office is also unsure how to deal with such applications.
As reported, in 2020, the copyright office rejected an application which listed
AI (RAGHAV) as the sole author for an artwork. However, a second application was
filed where a natural person and an AI (again (RAGHAV) were named as co-authors
for another artwork. This copyright office granted registration in this case.
The basis of grant of registration is not clear.
It appears that the Copyright Office granted this registration by mistake,
without applying its mind, as it issued a withdrawal notice, around one year
later. In the withdrawal notice, the Copyright Office shifts the burden on the
applicant to 'inform the Copyright Office about the legal status of the AI tool
Raghav Artificial Intelligence Painting App'.
The (human, co-author) applicant is now arguing that a copyright registration
cannot simply be 'withdrawn' once it has been granted. The procedure is a
rectification proceeding, to cancel the registration, before court. As reported,
this notice was issued around November 2021. The current status of this
proceeding is unclear. But as per the website of the Copyright Office, the
status of this application is still 'registered'.
Admittedly, this is old news. However, given that AI generated artwork has not
died out and has instead picked up momentum, this issue is still worth
Until this application is actually struck off the register (or corrected, if
such corrections are permissible), it may operate as precedent for subsequent
applications. While the Copyright Office may now believe that copyright
protection is conditioned on human authorship, its hands are tied. If subsequent
applications are denied, applicants may approach courts citing wrongful
rejection. Either way, it looks like the Copyright Office may need to spend
time, money and resources fighting this issue before court. This could have been
avoided if the office had not flip-flopped in its decision in the first
It is surprising to see copyright registrations being handed out without
thought. This is especially concerning given that High Courts are now elevating
the status of copyright registrations from mere registrations (results of
non-mandatory, administrative processes) to bullet-proof licenses. the Gujarat
High Court recently held that certificates of registration from the Copyright
Office provide immunity from copyright infringement. As he explains, this
interpretation is incorrect.
On one hand, there is a lack of legal precedent as to whether or not AI can be
classified as an author. A Parliamentary Standing Committee has recommended
'revisiting of IPR legislations and implementing a strong IPR framework' in
order to 'extract benefits from AI'. However, the law itself remains unchanged
and unclear. On the other hand, it may be procedurally possible to secure
copyright registration naming AI as a co-author and also use the registration as
a shield against claims of copyright infringement. A worrisome paradox.
New technology is puzzling and countries world over are grappling with the same
issue. For instance, the US Copyright Office has categorically rejected such
applications on the grounds that a 'work meets legal and formal requirements of
copyright protection only if it is created by a human author'. The UK
Intellectual Property Office, on the other hand, understands the evolving nature
of computer-generated works and is taking a wait and watch approach.
So I think Indian authorities should probably plan out their approach to tackle
such issues rather than take hasty action and create precedents that can turn
out to be dangerous in the future.
Creative robots' creations, which may be copyrightable if generated by humans,
are now questioned due to artificial intelligence. This question makes us
reconsider uniqueness. Formalizing originality will help determine whether
digital creations deserve copyright protection. International originality
The Berne Convention helped standardize copyright protection. The Berne
Convention covers copyright protection's most crucial
requirement�originality�broadly. Countries can use their own originality
definitions because there is no universal definition. "Notwithstanding the
necessity for monopoly protection of intellectual producers to foster creativity
and authorship, unduly wide protection would impede, rather than advance, the
objective," the US legal system self-restricts.
The Copyright Act of 1976 left originality undefined, although that the
Constitution protects "useful arts". The judge must define uniqueness since a
minimal amount of innovation is not enough. Formal, objective criteria should
assess originality, we believe. Private property is protected under US copyright
law because of its worth, but evaluating value without considering creativity
and uniqueness is unfair. Aesthetic analysis may result from valuing value over
effort. US law addresses art aesthetics.
This article proposes a more formal, descriptive framework for artwork,
bypassing judges' aesthetic approach. Judges outlines the aesthetic approach:
"artistic uniqueness is not the same thing as the legal definition of
originality in the Copyright Act. Artistic uniqueness may lie in a detail,
subtlety, or tint too tiny for a judge to detect. Descriptive study of
expression and effort should establish artwork uniqueness.
Thus, the objective viewpoint encompasses the expression's external
relationship, the audience's perception, and the work's similarity to others. A
descriptive approach will help judges and juries set an unbiased originality
- Baraniuk, C. (2016) Computer paints 'new Rembrandt' after Old Works Analysis, BBC News. BBC. Available at: https://www.bbc.com/news/technology-35977315 (Accessed: October 16, 2022).
- Yanisky-Ravid, S. (1970) Generating rembrandt: Artificial intelligence, copyright, and accountability in the 3A era - the human-like authors are already here - a new model: Semantic scholar, undefined. Available at: https://www.semanticscholar.org/paper/Generating-Rembrandt%3A-Artificial-Intelligence%2C-and-Yanisky-Ravid/66c5e36fb0d127b58dc9f4d13d6051e7fb23d0e4 (Accessed: October 16, 2022).
- Digital Art Movement Overview (no date) The Art Story. Available at: https://www.theartstory.org/movement/digital-art/ (Accessed: October 16, 2022).
- WIPO Copyright Treaty(WCT)
- Understanding the significance of perception in AI (no date) INDIAai. Available at:
https://indiaai.gov.in/article/understanding-the-significance-of-perception-in-ai (Accessed: October 16, 2022).
- Naruto v. David John Slater, 888 F.3d 418, 420 (9th Cir. 2018)
- Jacopo Ciani, Learning from Monkeys: Authorship Issues arising from AI Technology, Universitas degli Studi di Torino, Italy
- Christies. Available at: https://www.christies.com/features/A-collaboration-between-two-artists-one-human-one-a-machine-93 (Accessed: October 16, 2022).
- Telstra Corporation Ltd v. Phone Directories Company Pty Ltd (2010) 194 FCR 142
- Berne Convention for the Protection of Literary and Artistic Works as amended September 28, 1978
- Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS)
- Infopaq International A/S v. Danske Dagbaldes Forening
- See, e.g., the discussion on patents in Aashish R. Karkhanis & Jenna L. Parenti, Toward an Automated First Impression on Patent Claim Validity:
Algorithmically Associating Claim Language with Specific Rules of Law, 19
STAN. TECH. L. REV. 196 (2016)
- Copyright Basics, U.S. COPYRIGHT OFF. 1
(Sept. 2017), https://www.copyright.gov/circs/circ01.pdf; see also National Copyright
Guidelines, Requirements for Copyright Protection, SMARTCOPYING,
35 U.S.C. �� 101�02; Jonathan R. Tung, Who Owns the Creation of an
Artificial Intelligence?, TECHNOLOGIST , http://blogs.findlaw.com/technologist/2016/08/who-owns-the-creation-of-an-artificial-intelligence.html
Minnesota Journal of Law, Science & Technology (no date) Minnesota Journal
of Law Science & Technology | University of Minnesota Law School. Available
at: https://scholarship.law.umn.edu/mjlst/ (Accessed: October 16, 2022).
Anna Gaca, "The World's First Songs Composed by Artificial Intelligence"
Are Neither First nor Entirely Artificial,
- ESPOSTI, M.D.E.G.L.I., LAGIOIA, F. and SARTOR, G. (2019) The use of
copyrighted works by AI systems: Art Works in the data mill: European Journal of
Risk Regulation, Cambridge Core. Cambridge University Press. Available at:
(Accessed: October 16, 2022).
- Autor, D., Mindell, D.A. and Reynolds, E.B. (2022) Why 'the future of AI is
the future of work', MIT Sloan. Available at: https://mitsloan.mit.edu/ideas-made-to-matter/why-future-ai-future-work
(Accessed: October 16, 2022).
- Legal, D. (no date) News, Deriş Legal. Available at: https://legal.deris.com/en/news/copyright-and-anti-piracy/289-india-copyright-office-recognizes-co-authorship-of-ai-for-the-copyrighted-work
(Accessed: October 17, 2022).