CAN AN ARTIFICIAL INTELLIGENCE TOOL BE THE HOLDER OF INTELLECTUAL PROPERTY RIGHTS? A CONTROVERSY WHOSE DEBATE HAS JUST BEGINNING
Edgardo Ettlin
¡Dime qué debo cantar,
oh, Algoritmo!
¡Sé que lo sabes mejor,
incluso, que yo mismo!
Por ejemplo, esta canción.
¿Qué algoritmo la parió?
Me pregunto si fui yo.
¿La elegiste o te eligió?
Jorge Drexler, ¡Oh, Algoritmo!
SUMMARY:
I. Introduction - II. Can an Artificial Intelligence system or tool create? -III. Can
a work made by an Artificial Intelligence instrument or tool be
considered original? - IV. Can the AI be considered as
“author”? - V. Can a work created by Artificial Intelligence be
protected by intellectual property rights? - VI. Who owns the
intellectual property rights of a creation made by artificial
intelligence? - VII. Who can be
claimed for the infringements noticed in works whose creation artificial
intelligence was used, or were created by artificial intelligence? -
VIII. By way of closure
I. Introduction
Classical Law conceives Intellectual Property as a
personality right, within the framework of property rights. Along these
lines, we have postulated that “above all, Intellectual Property as a product
of intelligence is something of the human”, and that intellectual property
rights are first generation fundamental human rights (1). Moreover,
if the different intellectual property rights are species of property right and
if this is by definition one of the main human rights, it would seem a logical
conclusion to affirm that intellectual property rights are among the
fundamental rights of an individual.
The
Constitution of the Oriental Republic of Uruguay in its article 33, as well as
art. 491 subp. 1 of the Uruguayan Civil Code, provide that the “author's
right ” will be protected by Law, as well as that the productions of
talent or ingenuity “are the property of their author ”. The
idea of an author almost instinctively evokes us to think of
a Human Being. The expression “nationals of countries of the
Union” that use, for example, arts. 2, 3 and 6.1 of the Paris
Convention for the Protection of Industrial Property (or Paris Convention),
arts. 3 to 5, 6bis to 9, 11 to 15 and concordant of the Berne Convention
for the Protection of Literary and Artistic Works (or Berne Convention), and
arts. 6 and 7 of the WIPO Copyright Treaty (WCT), basically refers to
individuals or people.
We usually associate intelligence and talent with
humanity. Any product or work of his talent and intelligence, whether
it is the creation of a natural person or a legal entity as an "ex
lege" unit of a group of natural persons, is by definition human.
But what would happen if that intelligence
or that talent were not human? What would happen if a work came from a
non-human - non-biological activity? How would we consider that creation?
It is not the purpose of these disquisitions to
analyze what would be the nature of the productions of biologically non-human
beings, such as animals or eventually aliens (if one day it is verified that
the latter really exist). We wonder how the authorship of a
non-human intelligence developed by technology, through a machine, a
program, an application or algorithms by way of example, should be
assumed. Technology that has become known as “Artificial Intelligence”. We
could define provisionally and for the sole purpose of this study (“omnia
definitio periculosa est”) Artificial Intelligence (hereinafter
also interchangeably “AI” ) as the process of
cognition, reasoning, determination, and action elaborated by technological
systems .
In principle, artificial intelligence is the result
of human programming and operates on models created by human beings. But
that one is advancing; many of these instruments are “learning”, training,
adapting, exchanging, reprogramming (the technological details would exceed the
scope of our work). Even in some cases they are evolving towards being able to
develop or design their own processes, such as technologically
self-controlled, self-perfecting and self-recreating thinking, solving and
acting, who could their own decisions. Under these conditions, AI
would no longer be conceivable only as a simulation or as an imitation of human
intelligence: we are referring to artificial intelligence as an intellectual
and behavioral capacity “per seipsa”, with its own initiative and no longer
linked to any instruction or intervention. human (even though it may have
certain instructions, commands or startup parameters of its own).
If an artificial intelligence instrument or tool
could know, think, solve and act by itself, it could be understood that it can
also elaborate, produce by itself. This idea
forces us to consider, in the field of intellectual property, certain
questions:
1. Can an AI system or tool “create”?
2. Can a work made by an Artificial Intelligence
instrument or tool be considered “original”?
3. Can the AI be considered as “author”?
4. Can a work created by AI be protected by
intellectual property rights?
5. Who owns the intellectual property rights of a
creation made by artificial intelligence?
6. Who can be claimed for the infringements noticed
in works whose creation was used artificial intelligence, or were created by
AI?
We will try to provide some answers, which will
necessarily be provisional, debatable and open in a problem that is just
beginning to take shape in the world of Law.
II. Can
an Artificial Intelligence system or tool create ?
To address this issue, we must distinguish whether
the work was created by artificial intelligence on its
initiative and without the need for instructions or descriptions, from the work
created with artificial intelligence as an extension of the
human will whose process and even its result (that is, that is, if it is
maintained or if it is intervened) do not cease to be alien to it to a greater
or lesser extent. In the first case there is no human intervention except
in the contribution of some concepts or ideas from which the artificial
intelligence tool develops the work. In the second, AI is always an
instrument at the service of human creation.
The position adopted for or against considering an
artificial intelligence system as “author” is, seeing in the specific case that
is analyzed:
a) if the artificial intelligence processes
through the intervention of a human being to create the work
(creation with AI), or;
b) if
the AI generates the work by itself and autonomously (creation by AI).
It all depends on analyzing how AI performs in the
inventive or creative process.
If we think that artificial intelligence is only an
instrument that works for the person or uses inputs introduced by the person,
who ultimately is the one that will select the product that is produced,
facilitating its creation, it is already an operational extension of the
conception of an individual and therefore, the author is a human being. To
the extent that the initiative and conception of the work is human and the AI
tool is used to facilitate the production process, the person can be
considered as the creator, regardless of whether or not the final work
generated was in its result modified by it.
Likewise, to the extent that it can be verified
that, although there was some activity independent of the artificial intelligence
system, there was also some human intervention and this factor was not absent
in some way (for example in the selection of devices, in the design and prior
determination tasks), such an intervention ultimately shows a human creator,
and it could be argued in principle that there are no reasons to depart from
classical norms and solutions (2).
However, this should not be taken as an immovable
criterion.
Because
despite this, it can be said that the fact that a human being offers the artificial
intelligence system the parameters or characteristics in which the work must be
carried out does not necessarily make the person the author, to the extent that
the ideas, inspirations, procedures, methods of operation or mathematical
concepts themselves are not protected by copyright (art. 9.2 of the Agreement
on Trade-Related Aspects of Intellectual Property Rights -indistinctly “TRIPS”-;
art. 2 of the WCT) (3).
In terms of literary artistic creation, there are
already various artificial intelligence tools and applications available to
everyone (in the case of “Midjourney”, “Wombo”, “Dall-e” and many other
instruments), which allow the creation of portraits, images, “comics” and
literary works, available today within everyone's reach, based on certain
ideas, criteria or commands given by the user. For example, to create a
painting a person indicates that it is from “a city”, "in futuristic
style", “in Gothic mood", “with a somber color palette” and “in the
style of drawings of the '70s”. The tool itself creates and places the
images, establishes and adds the colors, defines a design and from it a final
product emerges. Who is the creator in this case? The person who gave
the bases that the image should have? The artificial intelligence tool
that, using its resources and making its decisions, issued a final work?
Now, if the AI can create without human
participation or intervention in the process of generating the work, we should
think about reputing it as an author, regardless of its
non-humanity or not.
In our opinion, everything depends on how the case
and its circumstances are considered, and how important the human intervention
or the AI tool was in the result of the work.
In the
case “Tencent vs. Yingxung Technology Shanghai”, it was analyzed the
assumption of an AI program called “Dreamwriter” created by Tencent who writes
about Business and Economics, and who had produced an article, a report on the Shanghai
Stock Exchange on August 20, 2018, which had been copied by defendant Yingxung
on her website. Tencent sued Yingxung for reproducing that article without
his permission. The Judgment of the People's Court based in Shenzhen of
the Nanshan District, Guangdong (China), did not affirm that the work
created by AI had been made by the Dreamwriter tool, but that it would have
been the product of artificial intelligence developed by the team of Tencent
that acted as a tool for the human being: it was considered that the
disposition and selection of the creative team of the plaintiff in data entry,
the establishment of activation conditions, template and selection of corpus
styles in this case belonged to intellectual activities of people belonging to
the company that were directly related to the expression specific to the item
in question. The Court found that the article in question was a work
created by the general intelligence of multiple teams and multiple divisions of
labor chaired by the claimant, and its creation process derived from the
selection and custom arrangement of the creator that reflected the needs and
intentions of the claimant as a whole, for the product of which it took
external responsibility. Therefore, the article in question was considered
to be of the legal person, as a work created by the claimant. Which
indicated that the article in question should be taken as written by
her. Therefore, in the absence of evidence to the contrary, this court
determined that the item in question was the work of a legal entity created by
the plaintiff (4).
The United States Copyright Office (USCO)
recognized on September 15, 2022 as the author of the comic or graphic literary
work “Zarya of the Dawn” to Kris Kashtanova, not to the artificial intelligence
application “Midjourney” that she had used. Kashtanova had registered
"Zarya of the Dawn" as a work of visual arts, claiming that she was
the creator in terms of the concept (the adventure of a non-binary person who
searched different worlds for mental tools to manage their emotions and to
communicate with other beings, a story said to have been based -according to Kashtanova-
on stories from her grandmother), certain artistic decisions and indications
for Midjourney to create the images, although the final product and the pages
generated by this tool were not altered in their result by Kashtanova. However,
the USCO itself informed Kashtanova on October 28, 2022 that it would initiate
the procedure to cancel the registration, based on the fact that the use of
Midjourney for the generation of the images and for the work produced had not
been taken into account when deciding on the registration, and in keeping with
its long-applied policy by the agency of rejecting copyright on AI-generated
works (in this case, the images and plot of the comic).
For the time being, in the Chinese ruling and in
the US registration procedure cited, we find certain guidelines to keep in
mind:
a) To the extent that the human being intervened,
whether by creating the artificial intelligence tool, by establishing the
criteria that the creation should have, or by intervening on the result of
production (the human being could approve what is believed, but eventually it
could not be satisfied with what artificial intelligence designed, and could
undertake another creative process or establish other instructions or “prompts”),
if artificial intelligence is part of or is used to only give reality to the
human creative process by facilitating or processing the work, we could suggest
that the human being should be reputed as the creator and as the author, beyond
the means used for the execution. Where appropriate, it should be presumed
in these hypotheses that the person is similar to a patron of the work;
b) If the
human being gives the basic ideas, but all the rest of the process in terms of
creation and the result are left to artificial intelligence, we might think
that this would be the true creator. Because in any case, the
requirements, instructions or commands that the human being gives or proposes
to the AI could be taken as ideas, initiatives or inspirations that in
themselves do not define the author;
c) It can
be granted that, in a scenario in which artificial intelligence would
eventually act on its own initiative and without the need to be given
definitions by people, no longer being a mere data processor, it can
create. And if artificial intelligence can create, it can be considered an
“author”. In this alternative, we could maintain that, to the extent that
it could be proven that the Artificial Intelligence instrument did not require
prior initiatives, nor programming, parameters or prior instructions given by
human beings, using their own resources both in the initiative and in the
creative process, its creation is its own and therefore artificial intelligence
would be the author as such even if it were not a human being.
From the foregoing, we can suggest that in view of
these scenarios it is necessary to determine in each specific case:
1) if the
artificial intelligence tool was only used to execute a human creative process
(supposition “a”), based on human programming and whose process or results it
supervises. It is what we call an invention, a work or a “creation with AI”:
the person is owned by the creator, insofar as he only uses the
AI;
2) if the AI instrument worked by taking its own
initiatives or creative or inventive decisions, it could be thought that it
created it by itself (hypotheses “b” and “c”). It is what we call an “creation
by AI”;
3) It
remains to discuss what happens when the human being gives the instructions,
the parameters or elements that the invention or creation must have, or
occasionally even the programming, but the AI in the work establishes its own
sequences, selects the information and adopts its decisions in a “personal” way
for clothing. Defining who the creator is here is, in our opinion, the
object of the discussion that is taking place in the “Zarya of the Dawn”
case. It is a limbic zone between 1. and 2., insofar as the human being no
longer acts, works or resolves in the creative-creative process, but rather all
the inventive work is done by the system, although it may be satisfied with the
result. To maintain that in this situation human intervention was reduced
to give ideas, inspirations, procedures, From understanding that the human
initiative defined what the AI should do and leaves it “the hard work” even
though the latter can make its decisions, we find ourselves in alternative “1”,
or we should assimilate the human being as the principal (analogically to what
happens in the work made to order) of creation.
And in the event that it is admitted that there was
activity and labor and effort from both the person and the AI instrument in
the creation, would we be talking about a co-authorship or a co-participation
between the AI and the human being? We understand that it is right, and
that it should be recognized.
If we admit that AI could or can
create "works" (creation by AI), it would be
necessary to study whether this creation could or can also be reputed to be
original.
III. Can
a work made by an Artificial Intelligence instrument or tool be
considered original?
Originality
is often associated with the projection or reflection of a personality in a
work, and this is important in considering whether the creation should be
granted intellectual property protection. Thus, various decisions of the
Court of Justice of the European Union (C-145/10, C-161/17, C-683/17)
determined that the work should be considered “original” as long as it is a “creation
that reflects the personality of the author, manifesting his free and creative
decisions”. In a classic decision of the Supreme Court of the United
States, it has been asserted that the “sine qua non” of copyright is originality: “As
in The Trade-Mark Cases, the Court emphasized the creative component of
originality. It described copyright as being limited to “original intellectual
conceptions of the author,” 111 U.S., at 58, and stressed the importance of
requiring an author who accuses another of infringement to prove “the existence
of those facts of originality, of intellectual production, of thought, and
conception.” Id., at 59–60.” (5).
There is nothing that prevents us from considering
that a work created by an artificial intelligence system or tool made
without any human intervention can be original , as long as the
creative product has emerged from its own resources, processes and decisions.
In the
case “Tencent vs. Yingxung Technology Shanghai” already cited (6), it was analyzed that the article
created with the AI created intelligence program called “Dreamwriter” had
original content, although as we also saw, the acting Court held that the
article in question was a work created by the general intelligence of multiple
teams and multiple divisions of labor chaired by the plaintiff, a work of a
legal entity. We must specify that if this said Tribunal had understood
that the work was the creation of the aforementioned artificial intelligence
system, the logical consequence would have been one: the work created by
artificial intelligence would have been original.
IV. Can the AI be considered as “author”?
The concept of “author” is usually examined
in its natural and obvious sense (art. 18 of the Uruguayan Civil Code) in
relation to a human being. In Spain, for example, art. 5.1 of
Royal Legislative Decree 1/1996, Consolidated Text of the Intellectual Property
Law, defines as author " the natural person who creates any
literary, artistic or scientific work ", without prejudice to the
fact that intellectual protection may benefit legal persons (article 5.2 of the
same Consolidated Text). In the United Kingdom, the legislation itself
settles the discussion by preventing any computer product from being considered
“author”, while in the “Copyright, Design and Patents Act 1988” it is
considered as “in relation to a work, means the person who creates it”
(art. 9.1 of the same; see also art. 7.1 of the “UK Patents Act 1977”), and in
its art. 9.3 declares with respect to literary and artistic works that “In
the case of a literary, dramatic, musical or artistic work which 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”.
The Constitution of Uruguay protects the “intellectual
work, the right of the author, the inventor or the artist " (art.
33), and the Civil Code of our country provides that “The productions of
talent or ingenuity are the property of their author ” (art. 491 of
the Uruguayan Civil Code). However, our regulations do not establish what
should be considered “author”. The art. 5th of Law No. 9.739
establishes as “intellectual production” “All production in the
domain of intelligence”, and establishes as the owner of intellectual
property “The author...” (art. 7 of said Law). None of these
provisions makes it explicit whether that author or that intelligence should
necessarily be human. See that if we stick to very literal legal
interpretation criteria (art. 17 of the Civil Code), our legislation does not
establish distinctions and therefore, the interpreter should not distinguish
whether the “author” is a human being or an artificial intelligence, without
prejudice to the fact that in each case the interpreter, registrar or Court
must establish who of them would be the author.
Considering that there is no artificial
intelligence that operates without human intervention, or at least as long as
that situation remains, the answer should therefore be “no”: AI could not be
held as the author.
On the contrary, if we could recognize that
artificial intelligence can create by itself (see Sections I and II), we should
also conclude that it could be the author of the works it devises. We
could not truly say that its creation or invention would be a “work without an author”.
The
possibility that artificial intelligence can be considered “author”
implies imagining the possibility of having to recognize such a quality in
something non-human. The concepts of “author” must then be reformulated or
expanded legislatively, doctrinally and jurisprudentially, so as not only to
understand them in terms of "people" but also
"intelligence" (without discriminating whether it is human or
artificial). Let's add that whoever creates is exercising will, damn it is something non-human. It
would no longer be a mere “thing” or “instrument”, but something more.
We wonder if we could eventually
conceive of artificial intelligence as a co-author, or as a collaborator, to
the extent that it has contributed its resources to human work. The answer
can admit two positions:
a) If AI
enhances and adds to human creativity with its own processing mechanisms, that
is, with its own intelligence even if it is not human, a new definition of “authorship”
could encourage the consideration of artificial intelligence as a “co-author”
or “collaborator” with the human being. Because there was a contribution from
both the person and the AI instrument in the creation, we are facing a
community of authorship;
b) If it is argued that there is already a human being, there is the intervention of his intelligence and ultimately, he is responsible for the creative process, the author would be the human person exclusively. The same could be said if it is believed that artificial intelligence cannot be considered “author” and by default, the authorship would be attributed to the human being involved.
We must bear in mind that if the AI produced the invention or work, whether at the initiative of a human being or of the AI alone, the AI should be considered the author if it is determined that it was the person who established or designed the content (we insist that the ideas, parameters, inspirations or suggestions of the human being user do not imply authorship). When the human being decides to adapt or modify that content, it must be analyzed in each case if we are dealing with an elaboration made with AI (which would admit human authorship), or if we are dealing with a generation by AI (with co-participation but without total authorship of the person); being the criterion to differentiate one or another hypothesis, the degree and quality of the human intervention on the creation of the AI.
It is questioned whether the work or invention created by AI would belong to the person who created the AI tool. In truth, we believe that it does not belong to him if it can be seen that he has tacitly or explicitly granted the use of the tool, freely or onerously, for third parties (without prejudice to his eventual liability as the creator of the AI, for improper use of third parties, if applicable). In this situation, the creator of the AI only makes the resource available, and does not participate in the creative process established between the third party and the AI, nor in the initiative of the third party or of the AI itself. However, when the human designer of the AI used his own AI tool to create, in that case we must think that he designed it for the creative process and would be considered the author of the work, unless the AI was created to act without the initiative or indications from him.
In the event that it is supossed that the artificial intelligence instrument could be considered “author”, “co-author” “collaborator” of a reputable work as original, one might wonder if said creation could be protected by intellectual property rights.
V. Can a
work created by Artificial Intelligence be protected by intellectual property
rights?
It is one thing to determine whether in the work
done by artificial intelligence, the artificial intelligence tool can be
considered “creator” and “author”. Both alternatives are possible.
Another is whether AI can be protected by
intellectual property rights.
According to the United States Copyright Office
(USCO-) the answer is “no”. In a request for registration of the graphic
work “A Recent Entrance to Paradise” made by the artificial intelligence
instrument “Creativity Machine”, it was rejected as an author and that registration
was denied by decisions dated August 12, 2019 and 14 February, 2022. The
registration request had actually been filed by a human being, Stephen Thaler,
but he was not listed as the author on the request. The central argument
given by the USCO to deny the copyright to Creativity Machine was that there
was no human authorship in the work “A Recent Entrance to Paradise”; although
Thaler tried to argue, without success,
The challenge that Doctor Thaler is posing around
the world is very interesting: trying to get artificial intelligence
instruments recognized as creators in order to be reputed as holders of
copyrights and patents (the issue could also be raised with trademarks), since
Thaler himself does not propose himself to the registers as author or inventor.
Greater discussion and debate has been sparked by
Stephen Thaler when he proposed patenting a “Food container and devices and
methods to attract greater attention”, under the authorship of the artificial
intelligence system “DABUS” (“Device for the Autonomous Bootstrapping of
Unified Sentience”) and figuring said AI instrument as the inventor. The
request had dissimilar registry and jurisprudential pronouncements in various
countries.
In
Australia, the Federal Court Judge Jonathan Beach found that the artificial
intelligence DABUS could be considered as “inventor” under the Australian patent
regime, in a judgment of 30 July, 2021 (7). The decision was based on
questioning why they cannot create our creations; and if they can create, “ergo”
they can be considered authors and protected as such. However, in second
instance the ruling of April 13, 2022 of the Full Federal Court of Australia revoked
Judge Beach's groundbreaking ruling, declaring that " The origin
of the right holder is a human entrepreneur " (8). Finally, the denial of the patent
was ultimately upheld on November 11, 2022 by the High Court of that country
(9). This last Body stated that DABUS
could not be considered an “inventor” under the Australian law for the
purpose of applying for a patent, and in the particular case, neither the
applicant nor any human who could have been identified was proposing himself as
an inventor. It is necessary to
highlight that Section 15.1 of the Australian Patent Law mentions that the
patent can only be “granted to a person”.
The United States Patent and Trademark Office has
the criteria of denying the intellectual protection of works created without
human authorship, which has been confirmed by Courts of Justice in that country
(10).
Likewise, in the United Kingdom, the Court of
Appeal of England and Wales ultimately decided that the DABUS System cannot be
held as the inventor, through its ruling of September 21, 2021 (11). The
case is nowadays before the Supreme Court of the United Kingdom, and there is
currently no resolution.
The European Patent Office also rejected the
registration of DABUS as the author of the invention in the final decision of
its Board of Appeals of December 21, 2021 (background J 0008/20), establishing
that an inventor must be a human being and not a machine, and the fact that
that machine was given a name was not a sufficient requirement for the
registration claim (12).
In a completely
opposite criterion, on June 24, 2021, the South African Companies and
Intellectual Property Commission granted the patent to DABUS and in favor of said artificial intelligence for
the invention of the food container that was unsuccessfully attempted to be
registered in other countries (13). This admission represents an
important precedent, which constitutes a vanguard in the subject of
intellectual property protection for artificial intelligence.
It is necessary to mention that
the position that admits the registration or the protection of the intellectual
property of works in which artificial intelligence is considered the author, is
reinventing the concept of “author”, according to new readings of that term and
the reality of a technology that becomes independent of the human being, coming
to manifest its own will.
VI. Who
owns the intellectual property rights of a creation made by artificial
intelligence?
In the case of patents and
trademarks, only those who are recognized by the respective registration
authorities can be holders of such rights (arts. 4 to 6 of the Paris Convention
for the Protection of Industrial Property; arts. 3, 4, 15, 27 to 29 more
concordant of the AADPIC, and particular norms of each State, region or
international system) (14). In the particular of copyright, the real rights over them can be
exercised without the need for registration, as long as it can prove the
original or derivative ownership (arts. 5.2 and 15.1 of the Berne Convention and
art. 9.1 of the AADPIC, art. 1.1 of the WCT; in Uruguay see article 6, subparagraph
2 of Law No. 9.739).
Let us remember that until now, intellectual property is conceived as a
human right; this is how it is protected in the highest international
instruments of human rights, and in the Constitutions of the different
countries. This explains why legal culture, legislation, doctrine and
jurisprudence are still mostly reactive or reluctant to recognize intellectual
property rights to artificial intelligence systems or tools. We have seen
in the previous Sections that legislations usually consider a “person” as the
“author”, and that where this is not explicit, it is taken for granted.
That is why it is very difficult to conceive that
AI, by itself, can be the owner of intellectual property rights.
The only
way in which the intellectual property of AI could be protected would be to
give artificial intelligence legal clothing by granting it legal status or
considering it a center of imputation of rights and obligations (regarding this
last possibility, it would be similar to as provided with respect to the rights
of animals, which in Uruguay are subject to legal protection and guardianship -article
21 of the Civil Code, article 1 of Law No. 18.471, articles 27 and 98 of the
Decree of the Executive Power No. 62/014-). Although such fiction cannot
be sustainable, if artificial intelligence physical needs persons for expressing
their will.
Why do some begin to think, and others to propose in
practice like Stephen Thaler, that artificial intelligence instruments should
be recognized as the authors or inventors themselves? Because we
are already facing the reality that there are “things” that
are creating or can do so without the need for human intervention; and in
another reasoning, if the AI is in charge of a good part or almost all of the
creative process, its authorial work, or at least the co-authorship,
collaboration or creative credit that it has in the work, should be
noticed.
If it is intended to recognize intellectual
property rights protection for artificial intelligence, we still have legal
gaps to fill.
Of course the legislator, who usually works magic
with realities, could issue regulations recognizing ownership or special nuclei
of intellectual rights, or their protection of some kind, eventually moral, to
artificial intelligence tools. The Courts could also lead progress in this
regard, taking precedence over what the Laws define. But both the
legislation and the jurisprudence find the stumbling block of how they can give
protection to those who are not people because they are not subjects of law:
artificial intelligence or artificial intelligence tools are not human beings,
they are strictly things, they cannot be protected; we got to this point.
The situation that denies intellectual protection to artificial
intelligence, based argumentatively on the fact that the artificial intelligence system cannot be protected
as an author because it is not a human being, leaves the work created by it
unprotected, when no human creator appears as headline. Anyone could, in
this case, plagiarize it, adapt it or (in countries where it is not allowed
without the authorization of the owner, as in Uruguay -arts. 35 and 44.A.4 of
Law No. 9.739-, Argentina -arts. 2°, 25, 26 and 36 Law No. 11.723- and
Dominican Republic - articles 6º num. 1º and 11 Law 65-00- for example) even
parodying it (this last case typical for literary and artistic works).
Likewise, no human being could claim the
intellectual property of an invention or creation made by AI,
nor claim any paternity, for not being its author. If a person claimed
that, then they should postulate that they made the work using AI or with AI,
the person having intervened and guided the creative process; although in
this last alternative we are already talking about a human creation, not an
invention by AI.
With such a perspective, the illegal use or
reproduction of a work made by artificial intelligence, if it cannot be
protected, we will only have an ethereal moral reproach (we are not talking
about moral intellectual property rights, but simply about issues of a moral
nature and not legal) and not a legal guardianship.
Would then be the works or inventions generated by
AI that are not considered authorship or human co-participation, “res nullius”? Would
we find ourselves in these circumstances facing a gap or a legal void, where
the protection of intellectual property would not reach or where it would not
be feasible or possible?
It is clear that with respect to trademarks or
patents and their related rights, if they cannot be registered in the name of
an AI system or instrument, there will be no way to protect them
intellectually.
With the literary and artistic works created by AI,
it is also obvious that they cannot be the object of private economic or moral
rights of intellectual property. Would it be feasible to apply when not
having human owners, the works made by AI could belong to the public domain of the State
due to copyright and related rights? A negative response is also required,
because even when the AI is considered to be an “author”, it cannot be the
owner of intellectual property rights and, therefore, its works lack such
protection or potential for “erga omnes” opposability, not having
possibility of passing at any time copyright or similar rights, to the State.
Despite all this, let us not forget that it is
still a general principle of law that unauthorized use of the efforts or
creations of others is improper, and therefore should not be allowed. Although
this “alien” is a machine, programming, data sequencing or algorithms, we are
talking about the same thing.
VII. Who can be claimed for the infringements noticed in works whose creation
artificial intelligence was used, or were created by artificial intelligence?
How can
the creators of protected works claim for their intellectual property rights,
if their contents or works were used by artificial intelligence? Against
whom can a claim be made if in the work made by artificial intelligence, the
machine, instrument or algorithm used repositories, data or resources of
creations or related to creations eventually supervised? How can it be
legally addressed when a subject takes improper advantage of the creation made
by AI?
In a
similar way to what happens in the general system, it should be studied as a
matter of fact if only ideas, inspirations or stylistic resources were taken by
the AI, or if this use came to adopt expressions or content adapted, parodied
or plagiarized without authorization of the owners of the protected contents or
creations.
If the work was generated by AI without any human
intervention, whether it is recognized as an intellectual property owner or
not, there is no possibility of asserting third-party rights against it, because
artificial intelligence is not human and therefore cannot be subject to of law
subject to obligations, nor can it be legally represented.
The question will then be: “Cherchez l'être
humain”.
We can answer like this, openly and without
prejudice, according to these possibilities:
a) If it is admitted that the author of the
questioned work is artificial intelligence itself without the intervention of
any person, or if it acted by itself and on its own initiative, since it is not
a subject of law (art. 21 of the Uruguayan Civil Code), the tool obviously
lacks passive legitimation. However, if artificial intelligence eventually
acted on its own initiative, the human being or beings that created it may be
sued, based on the fact of having designed it or having used or used it (in
Uruguay, arts. 1319 and 1324 of the Uruguayan Civil Code);
b) If it
is understood that the work in question was authored by a human being who used
or disposed of the artificial intelligence tool as a means to carry out the
creation process or as an extension of its activity (in this eventuality the
person is being considerable as author), the owner of the protected rights may
sue him for liability (both for his own fact and for the fact of the thing used
or that has been used or that is used; articles 1319 and 1324 of the Uruguayan Civil
Code);
c) If in
the process of the faction of the work by artificial intelligence there was any
human intervention, of any kind (since the invention of the artificial
intelligence tool, as well as in the initiative, in the instructions, commands
or parameters, as in the final selection), the owner of the protected work can
sue the individual regarding his participation or co-participation (both by his
own fact and by the fact of the thing used or that has been used or that is
used; arts. 1319, 1324 and 1331 of the Uruguayan Civil Code) even if it is
considered that the creator is the AI;
d) In
hypotheses "b" and "c", you can sue jointly (eventually
"in solidum" or jointly) both the creator of the AI tool and the
human being who was involved in the creation process in which the tool acted;
e)
Discussions may arise as to whether the person who created the artificial
intelligence tool made it freely available to the public (in which case they
would not have gotten rid of its legal or material disposition), if he did so
under a contract granting free or onerous license (not losing the provision on
the AI in terms of the economic benefit that the licensee gives to grant the
licences), or if he transferred its property free of charge or for a fee to a
third party (in which case it lost the provision on the AI
instrument). What will make the creator of artificial intelligence
liable in the first case for his own act or as a guarantee, in the second as
responsible as a guarantee, and will be exonerated from reproach in the last.
In the event that a third party or third parties
have adapted, parodied or plagiarized a work, invention or design of any kind
produced by artificial intelligence, only those who can justify any
registration, right or ownership of the creation may claim them for the
corresponding property rights made with or by AI,
according to the general rules. Whoever intends to apply as a legitimized
asset cannot claim a representation of the AI tool because things cannot be
represented, unless express legal dispensation or that a legal legal fiction
has granted the AI some protection as a center of imputation rights and
obligations; to be it evaluated according to the legal order of each
country or region.
We could
imagine, in another aspect, that the AI could just act in such a way that it
was trained or could train itself to recognize, or to prevent not using,
certain content that could be protected by intellectual property
rights. This solution could contribute, without the need to resort to
summons, claims or litigation, to an adequate balance of the legal assets at
stake.
It remains to study how criminal responsibility
would be enforced.
In the case of works made with artificial
intelligence, as soon as intellectual property rights whose owners were human
beings can be brandished, the general rules apply.
Against those who illegitimately use creations
designed by artificial intelligence systems, in the works
created by AI there are no, at least until express legislative
reforms are verified, owners or rightful holders of intellectual
property. Will the offenders of these latest inventions go unpunished in
this hypothesis, or outside of the criminal punitive claim? The answer
seems positive, because there is no human intellectual property attacked.
VIII. By way of closure
New technologies and advances in informatics have
generated non-human intelligence (machines, programs, applications or
artificial intelligence algorithms) with the ability to create works by
themselves and without the need for human intervention or control. These
creations can in certain cases, be noticed as original.
It is necessary, then, to question whether
artificial intelligence has the capacity without the need for human
participation (even if it eventually processes the data with certain commands
or guidelines given by it) to create with originality, whether artificial
intelligence instruments should be recognized as true authors. The second stage
is to consider whether AI should be recognized no longer with a credit or a
conceptual participation in the invention, but with the entire protection of
its activity through intellectual property rights. In these circumstances,
the term “author” must be redesigned or resignified, in accordance with this
imminent reality.
Determining how this AI creativity should be
regulated and protected from Intellectual Property Law is currently a problem
that is knocking on the doors of legislation, registries and Courts. Although
for the moment the answer is mostly negative in terms of granting artificial
intelligence the protection of intellectual property over its creations, such a
reality must one day be addressed and it will have to be analyzed whether it
should not be protected affirmatively.
In the meantime, the Courts and registries, with
the legal instruments available to them in their respective countries or
regions, need to solve with ingenuity and common sense the issues that involve
eventual intellectual property claims of artificial intelligence or problems
related to attribution of interests in the creation by AI, balancing the
different rights at stake within the applicable legal order to find solutions,
which may not be unanimous, to recognize and give the treatment that
corresponds to these new realities.
The controversy on the matter is open and like
Leuconoe's Response, it offers us Space (15)
towards what is to come. But what will come next?
Propius Tibi, Domine.
Die fifteenth mensis Januarii, Anno Domini MMXXIII
[1] ETTLIN
Edgardo, Estudios sobre Justicia y Propiedad Intelectual, La Ley Uruguay
- Thomson Reuters, Montevideo, 2021, pp. 1 and 3.
[2] Judgment of the Beijing Intellectual Property
Court of April 2, 2020, case “Gao Yang v. Youku”.
[3] In Uruguay, the AADPIC was ratified by
National Law No. 16.671, and the WCT was ratified by National Law No. 18.036. The
Paris and Berne Conventions were ratified by Decree-Law No. 14.910.
[4] Decision of 12.24.2019 of the Nanshan
District People's Court, Shenzhen, Guangdong Province (People's Republic of
China), case “Tencent v. Yingxung Technology Shanghai”.
[5] Supreme Court USA -
Feist Pubs., Inc. v. Rural Tel. Svc. Co., Inc., 499 US 340 (1991).
[6] See Section II.
[7] Thaler v Commissioner of Patents [2021] FCA
879.
[8] Commissioner of Patents v. Thaler [2022]
FCAFC 62.
[9] Thaler v Commissioner of Patents [2022]
HCATrans 199.
[10] Judgment of August 5, 2022, case “Thaler
v. Vidal”, United States Court of Appeals for the Federal
Circuit. Judgment of September 2, 2021, case “Thaler v. Hirschfeld”, United
States District Court for the Eastern District of Virginia.
[11] Thaler v Comptroller General of Patents Trade
Marks And Designs [2021] EWCA Civ 1374 (21 September 2021). Art. 7.1 of
the “UK Patents Act 1977”: “Any personmay make an application
for a patent either alone or jointly with another. ” (highlighting
ours).
[12] https://www.epo.org/law-practice/case-law-appeals/recent/j200008eu1.html
(accessed 14.1.2023).
[13] https://ipwatchdog.com/wp-content/uploads/2021/07/AP7471ZA00-Notice-of-Acceptance-1.pdf
(accessed 14.1.2023).
[14] See Note 3.
[15] RODÓ José Enrique, Motivos
de Proteo second edition, Editorial Cervantes, Valencia, 1918, pp. 40-41.
