Rights to the Result of Intellectual Activity of Artificial Intelligence in Terms of the Dualism of Intellectual Law

Akhramkina Kira Anatol'evna ORCID: 0000-0002-1224-5406 Postgraduate Student, Russian State Academy of Intellectual Property 55a Miklukho-Maklaya str., Moscow, 117279, Russia kirra4@yandex.ru Other publications by this author

The purpose of this work is to determine the legal affiliation of the results of intellectual activity of artificial intelligence (hereinafter – AI) among the elements of the legal relationship necessary to build a model of legal regulation related to the adaptation of Russian intellectual property law to new realities and challenges associated with the development of AI systems. The significance of the study lies in the fact that currently there is no clear understanding in the Russian and world legal system about the regulation of legal relations involving AI in general. In addition, we share the point of view of M.A. Efremova and T.M. Lopatina that the issue of granting a robot with AI the status of a subject in civil law in general seems to be primary in relation to considering it as a subject of criminal law relations or other legal relations in other branches of law, and, therefore, in this communication is a determining and important factor and should be comprehensively considered^[1]. Modern digital technologies are developing much faster than changes in the legal framework associated with it. In this aspect, intellectual property law, as a branch of law directly related to innovation processes, faces the greatest difficulties. Currently, the federal project "Artificial Intelligence" is being implemented in our country, the task of which is to create conditions for enterprises and citizens to use products and services based primarily on domestic artificial intelligence technologies. Also, by Decree of the President of the Russian Federation No. 490 dated 10.10.2019, the National Strategy (hereinafter referred to as the Strategy) for the Development of Artificial Intelligence for the period up to 2030 was approved.

Among the most significant legislative projects in this area should be highlighted the draft federal law "On amendments to the Civil Code of the Russian Federation in terms of improving the legal regulation of relations in the field of robotics" and "Model Convention on Robotics and Artificial Intelligence", developed by the Research Center for the Regulation of Robotics and Artificial Intelligence. Both of them, based on the principles of security, loyalty and consideration, are aimed at building a system of legal norms regulating the relations of AI and robotics in the social sphere. The ideas proposed in them on the creation of a Unified state Register of Robotics, on the application of the norms of legal responsibility to the owner and user ^[2],[3] are not uninteresting.

In accordance with the purpose of the study, the tasks are set: to define the concept of "artificial intelligence", to consider AI from the point of view of the subject of legal relations developing in the field of intellectual property, to substantiate the point of view of the existence of property and non-property rights, participants in legal relations in the field of AI inventions, from the point of view of the dualism of intellectual property rights.

The article analyzes approaches to determining the place of AI in the structure of legal relations. Some authors refer AI to a variety of objects of legal regulation; others consider AI as a specific subject of law, in the latter case, the definition of rights and obligations is difficult, due to both the diversity of AI systems and the definition of the role of developers, operators, etc.

The importance of AI in the modern world is huge. Starting with the invention of the transistor and integrated circuits, it has become a part of human life (Google translator, voice assistants, contextual advertising). In fact, he is part of a process called "transition" and "transformation", which means the birth of a new understanding, including human nature. Not so long ago, the testimony of a robot was documented in a Swiss court when considering a case of a collision between a sports car and a scooter, as a result of which the driver of the scooter was seriously injured. The court took into account the indications of the car driver's drowsiness detection system and concluded that he was guilty. Before the collision, the system warned the driver several times about suspected drowsiness. In fact, the car can monitor its driver. It is assumed that in the future, humans and robots will be so closely related to each other that it will be difficult to establish cause–and-effect relationships, and our current understanding of responsibility will lose relevance^[4]. The modern French system of legal regulation already makes extensive use of digital systems introduced into justice, to facilitate the collection of some statistical information and analysis, aimed at further elaboration of the legal structure of responsibility and the distribution of intellectual property rights ^{[5; p.51]}. The services introduced with the help of AI are undergoing changes, being refined and adapted, a number of previously provided services have already been changed (automated decision–making in online arbitration and the official publication of judges' analytics, the so-called "predicted justice", have been excluded). Debates about AI are very often translated into consideration of legal issues about human-robot interaction with AI, based on the ideas of both cooperation by robots and granting them rights similar to or equal to human rights.

Currently, there is no unambiguous definition of the concepts of "thinking", "consciousness", "intelligence" in scientific discourse. The content of these concepts depends on the context in which they can be used, or on the paradigm in which they can be considered. According to G.I. Kolesnikova, the difference between humans and AI is mainly in motivation: "It is motivation that gives quality to actions. Since the ancient world, there have been three types of motivation: two of them denounce a slave (an action out of fear of punishment or a desire for encouragement) and one free person (an action based on one's own beliefs)"^{[6; p.37]}. In our opinion, motivation can give quality to actions, but it is unreasonable to consider a direct connection between AI motivation and the quality of its actions. In addition, it would be more correct to judge goal-setting as a whole as a conscious determination of one's needs and motives, and not just as an incentive to action, since goal-setting is a broader concept, including including not only setting goals, but also planning to achieve them. In addition, from our point of view, the use of AI instead of humans in work matters is possible with 100% accuracy in predicting the result, which is not always achievable.

American scientists note: "What AI and machine learning are really good at is things that are finite and controlled, limited tasks, but they don't know how to cope with ambiguity. Machine learning and AI are inherently deceptive, they implement the models produced for them. These are probabilistic reasoning, they make decisions based on the best available information, and, as a result, sometimes they may not be accurate"^[7]. AI "does not have the mind to complete the mental act necessary for invention or discovery^[8]. The European Commission defines AI as a system "demonstrating intelligent behavior. These systems analyze their environment and take measures with a certain degree of autonomy to achieve certain goals. Systems based on artificial intelligence can be software and work in the virtual world (for example, voice assistants, search engines, speech and facial recognition systems) or embedded in hardware devices (for example, robots, autonomous machines, drones, etc.)"^[9].

The disclosure of the term "artificial intelligence" in the Strategy is carried out through "a set of technological solutions that allows simulating human cognitive functions (including self-learning and searching for solutions without a predetermined algorithm) and obtaining results comparable, at least, with the results of human intellectual activity when performing specific tasks"^[10]. The technologies and technological solutions in which AI is a mandatory element include robotics and management of unmanned vehicles.

D.A. Manoshin defines artificial intelligence as "a set of software technologies designed to implement actions identical to those performed by the human brain, including independent decision-making. There are a number of specialized platforms for creating applications based on artificial intelligence. They are based on the principles of artificial neural networks that are able to accurately recognize speech and objects, as well as independently generate actions necessary for the functionality of the software" ^{[11; p.21]}.

In our opinion, a brief but precise definition of AI is its vision as "a group of algorithms that have the ability to modify themselves based on the data entered"^[12]. E.V. Kupchina notes the key feature of AI – the ability to self-study: "Analyzing the data obtained from their own experience ... is able to solve many analytical problems, such as how: evaluation, classification, forecasting and clustering of data"^{[13; p.50]}. P.M. Morkhat by artificial intelligence technologies means computer or cyberphysical systems with anthropomorphic (human-like) "intelligence", while defining artificial intelligence as "fully or partially autonomous self–organizing (and self-organizing) computer hardware and software virtual (virtual) or cyber–physical (cyber-physical), including bio–cybernetic (bio–cybernetic), a system (unit) that is not alive in the biological sense of this concept, with appropriate mathematical support, endowed/possessing software–synthesized (emulated) abilities and capabilities ^{[14; p.5],[14; p.30]}. Within the framework of this detailed definition, it is necessary to point out the difference between a number of concepts that are often synonymous in the existential understanding. Automation and robotics can contain AI or include it, but this is optional. "Automation is simply when a machine replaces a task that a person performed, but it doesn't have to be a cognitive task. Robotics is the hardware part of the sphere, and, in fact, it acts or acts in the world to perform a task. Typically, when we talk about artificial intelligence, we are talking about algorithms, machine learning and/or natural language processing, and AI may include one or more of these components. By algorithms, we simply mean a series of steps to complete a task, except that these steps are performed by a computer and not by a human"^[15]. Not every robot is equipped with AI and not every AI is embodied in robotics. In addition, an important distinguishing feature of a "strong" AI in comparison with other technical solutions (video cameras, recorders) is the transformative function and interaction with the world, while the tasks of the latter are to perform the simplest functions and simple tasks, for example, fixing, performing mathematical calculations, etc.

May Lee notes that AI is not yet sufficiently developed to fully imitate human thinking, having passed the Turing test, which consists in assessing the computer's ability to demonstrate intellectual behavior equivalent to human ^[12]. S.A. Afanasyev argues that the human mind is also artificial, since it represents a deviation from the biological function, a derivative in relation to thinking^{[16; p. 76]}. He also touches on both the general question of the applicability and creation of legal norms regulating the relationship between man and non-human reason, based on equality, and the private one about the possibility of relying on another, extra-human reason, including when assessing human errors and influencing them ^{[16; p. 77]}. Marvin Minsky, a pioneer in artificial intelligence research, insisted that "the human brain is a computer made of meat"^[17]. A. Duflot asserts the principle of the priority of man and his interests over AI, mainly "in the field of the distribution of intellectual property rights" in relation to AI systems and the results of their activities"^{[5; p.51]}.

If we consider AI as an inventor, then Ernest Fok's statement is true that "people can give machines the task and materials necessary for an invention, but this does not make it an inventor. Invention requires contributions to the concept of the invention, and inventors can adopt ideas, suggestions, and materials from others as long as they retain intellectual dominance over the creation of the invention. Similarly, a programmer or user who directs an AI inventor to create inventions in this field cannot qualify as a co–inventor of the invention"^[8]. He notes that the US patent regime has the ability to promote innovation and change the patent policy itself, despite the fact that international regimes do not recognize computer technologies as inventors. Patent law, in his opinion, should focus on the nature of the invention itself, and not the subjective thought processes by which it was achieved. Obviously, such enthusiasm is based on the fact that the United States is essentially a pioneer in the registration of extraordinary scientific solutions: starting with the issuance of a patent on April 12, 1998 for the "Harvard Mouse", an animal born in a retort with the help of genetic engineering, ending with the receipt in 2005 by John Koza of a patent for an invention created by AI, an "inventive machine" that used genetic programming (modeled after evolutionary processes) to improve the control system without assistance (in the patent application, a human inventor was indicated, who actually did not participate in the invention) ^[8]. From the point of view of Russian legal realities, the products and results created by AI can theoretically be recognized as inventions and acquire protectability, according to Article 1350 of the Civil Code of the Russian Federation, if they have an inventive level, are industrially applicable and were not previously known from the state of the art. Because of this, it also seems to us that the non–patentability of an invention created by an AI program in offline mode is unjustified. Despite the fact that modern Russian legislation does not contain instructions regarding the regulation of this type of relationship, our conclusion does not contradict the list of intellectual property objects and their classification given by WIPO, where all types of objects protected by intellectual property rights are listed in seven paragraphs, and it is also noted: "Intellectual property includes rights relating to all other rights resulting from intellectual activity in industrial, scientific, literary or artistic fields"^[18].

Let's consider some points of view on artificial intelligence as a subject of law.

The main topics of discussions on the relative views on the legal status of AI are listed by I.R. Begishev: on assigning robots the legal status of "electronic persons", on recognizing AI as a full-fledged cybersubject of society, on the possibility of endowing virtual persons with legal personality, on endowing a particular system with the status of a subject of law, on recognizing the AI system as a subject of copyright and patent rights, giving the construction of a legal entity to a robot or AI^[19].

E.L. Zateyshchikova points out that artificial intelligence has the following characteristics of a legal entity: he is isolated externally, has a material form, is personified, acts as a single person, is capable of expressing his will through legally significant actions, but the question of his legal capacity is debatable due to his lack of consciousness and soul as moral qualities of a person ^{[20; pp. 380-381]}. She suggests the following approaches to solving the issue: to consider artificial intelligence as a technical means with a legal regime of a thing or to give it the status of an electronic person (as an analogue of a legal entity), in the second case, since the will as an integral attribute of the subject of law is not inherent in artificial intelligence, a legal fiction is actually applied^{[20; p.381]}.Simultaneous recognition of AI as an object and a subject of civil law will give rise to absurd "legal relations". Modeling such a possibility, V. Vitko cites the example of a person receiving a claim from mold "in his apartment, demanding protection of her rights in court, who considered herself the author of an image on the wall modified by a person in which, in her opinion, her "feeling" was expressed^[21].

F.V. Uzhov reveals the concept of an "electronic person", introduced applicable to AI by the European Union, as a carrier of artificial intelligence (machine, robot, program), possessing a mind similar to a human one, the ability to make conscious and not based on the decision algorithm laid down by the creator of such a machine, robot, program, and therefore endowed with certain rights and responsibilities and distinguishes among its rights the right of the carrier to inviolability (according to which all changes must be legitimate and authorized by the appropriate decision of the authority) and the right to authorship (according to which the AI should the electronic person should own the right to both the creation itself and the copyright to it)^[22]. In our opinion, this point of view has novelty, but the ways of implementing the author's idea are unclear. In addition, the implementation of the task of civil law in the form of regulation of property and some non-property relations is possible without endowing AI with legal personality.

A.D. Pomeranets and V.V. Kolentsova believe that at the moment of the development of AI technologies, endowing AI with legal personality is not necessary, but there is a need to "create an electronic veil, which can be used as an attempt to limit the liability of the owners or owners of this technology"^{[23; p. 275.]}. An example should be Elon Musk's statement in connection with the publication in the media of information about accidents involving autopiloted Tesla cars. He stated that manufacturing companies should not be held responsible for accidents that a self–driving car gets into, as an elevator manufacturing company should not be held responsible in every case of an elevator stop. Based on this, the property responsibility should be borne by the owner of the source of increased danger, and not by the manufacturing company.

V. A. Shestak and A. G. Volevodz also analyze the idea of endowing artificial intelligence with legal personality and indicate the following signs of the "reasonableness" of robots: the ability to analyze data; the ability to adapt their behavior; autonomy and the possibility of self-learning^[24].

As A. Gurko notes, since AI is not a person, it is impossible to recognize personal non-property rights to the works of artificial intelligence^[25]. That is, in fact, they deny recognition of non–property rights for anyone in this area of legal relations.

English scientists believe that the recognition of legal personality is not effective for the following reasons. Firstly, it is connected with the concept of legal personality in general, its optional relationship with the concept of personality, "legal personality" in this sense is represented by the endowment of the legal system of the object with certain rights and obligations. Secondly, legal personality is not universal: the scope of rights and obligations of different subjects is different. Thirdly, the granting of legal personality is not equivalent to the effective interaction of the subject with the legal system as a whole ^[26].

From our point of view, the rights to artificial intelligence inventions should be considered from the position of dualism in intellectual property law. Speaking about the concept of dualism in general, we are based on the opinion of G.E. Zborovsky that the theories, teachings, concepts of modernity are epistemologically shifted relative to T. Kuhn's theory and are not of a revolutionary scientific nature^[27]. The concept of dualism of intellectual property seems to us to be a speculative system, an understanding according to which moral and property rights are presented separately. "Formally, we can talk about dualism in copyright when the institute of moral rights is included in the copyright law. Naturally, a clear distinction between moral and property rights should be preserved"^{[28.; p. 66]}.

If we talk about modeling the legal personality of AI by analogy with a legal entity, which is a property complex, which is granted legal personality, then such a fiction, which generates the endowment of a legal entity with rights and obligations, can be applied to AI, but only taking into account the limitation of rights and obligations in comparison with a person. However, the recognition of AI as a subject of law in legal relations related to intellectual property, by analogy with a legal entity, contradicts the Charter of Authors' Rights (adopted on September 26, 1956 in Hamburg at the 19th Congress of the International Confederation of Societies of Authors and Composers (CISAC) (ch. 2, paragraphs 5 and 6): "5. The basis of copyright is an act of intellectual creativity. Its source lies in the very nature of things. The law should only regulate this activity, it should never depend on legal formalities. 6. The ownership of copyright is based on an act of intellectual creativity, and only an individual can have the right of authorship to his creations. A legal entity should never be considered as the original copyright holder of an intellectual work. It should be rejected as unacceptable the concept according to which the author is considered a simple employee of an industrial enterprise, to whom the work created by the author would belong as an ordinary product of this enterprise"^[29].

In our opinion, the limitation of the legal personality of AI, firstly, is based on the fact that artificial intelligence is a machine that generalizes information already available to people about the surrounding reality, it can create something similar, not devoid of novelty, but relying on what has already been created by people, and a person is able to create an original work with zero. In this regard, the most significant are the beliefs of V.N. Sinelnikova and O.V. Revinsky that "if a program capable of creating new programs is developed, then the rights to these new programs will be with the developer of the original program or with the one to whom he transferred his exclusive right to it," since "the result of the work "artificial intelligence" is the result of the intellectual activity of the human creator who created this "artificial intelligence" ^{[30, p.22],[30, p.23]}.

The conclusion of the Resolution adopted by the participants of the round table "Intellectual Rights and protectability of objects created using artificial intelligence in the digital age", held within the framework of the IX Perm Congress of Legal Scientists (2018), that "works created by artificial intelligence are not currently objects of copyright, since it is unreasonable to recognize as their creator the person/persons who created the corresponding computer programs" seems to us unconvincing on the following grounds ^{[31; p. 163]}. Works created by artificial intelligence are not actually classified as objects of copyright at the moment, but this does not mean that they will not be affected by new legislation in the near future, since some results of the functioning of algorithms do not differ from works created by man and such results should be granted legal protection. In addition, the programmer's participation consists in using a set of procedures, techniques and methods to train the machine to create results similar to copyright objects, actually at the expense of the created program. If we take as a basis the understanding that the program is a step-by-step command, the programmer's instructions for the execution of the machine, and the AI is a complex of these commands, then it cannot be denied that the works created by AI appeared absolutely without human participation. The protectability of intellectual property objects directly depends on their connection with the personality of the author. And the developer's task is undoubtedly creative, it is not limited to writing code, it also includes understanding the task, translating it into a technical statement, describing the requirement, understandable AI, requires human experience, the ability to communicate, think logically, make assumptions, draw conclusions, etc. Even if, in comparison with the work of the artist, creativity is not so obvious here, and the role of the author approaches the role of the editor, but his minimal creative contribution is obvious. In addition, the connection between the author and the functioning of the AI system is obvious from the point of view of his responsibility. We adhere to the opinion set out in the PACE recommendations "Merging with Technologies, Artificial Intelligence and Human Rights" dated April 28, 2017 No. 2102 states that it lies with a person regardless of the circumstances of what happened, and even references to the independence of decisions made by artificial intelligence units cannot exempt their creators, owners and operators from responsibility ^[32]. It should be clarified here that the grounds for liability may arise as a result of the actions of persons other than the author (the manufacturer, the person who carried out the maintenance, etc.).

In addition, according to Article 1257 of the Civil Code of the Russian Federation, it does not establish requirements for the amount of creative contribution for the recognition of authorship. In fact, this means that "if a programmer makes a relatively small creative contribution to the creation of a work with the help of artificial intelligence, he can be considered as its author"^{[33; p. 41]}.

Secondly, our point is mediated by the specifics of AI work from the point of view of human physiology: an artificial neural network is built on the principle of functioning of nerve cells of a living organism, but it is inferior to the structure of a biological neural network in terms of the number of layers of neurons, and is also characterized by sequential activation of layers in an artificial network, while in the human brain the exchange of information between neurons are a more complex process, parallel and asynchronous^[34].

As for the exclusive rights to such products, the initial attribution of them to the author or rightholder of AI seems reasonable to us for the following reasons: if the subject of law is the rightholder of the neural network, then the results of its activities belong to him. Here we can give an example of an analogy with the Roman right to fruits: "The fruits, from the moment of separation from the fruiting thing (separatio), i.e. from the moment from which the fruits become a separate thing, belonged only to the owner of the latter"^[35].

In the study of E.A. Mikhovich, the following options for granting exclusive rights to a work created by AI are considered:

"1. An option in which the exclusive rights to a work created with the help of AI are transferred to artificial intelligence.

2. A variant of hybrid authorship, in which the exclusive rights to the work are distributed between artificial intelligence and humans.

3. An option in which the exclusive rights to a work created with the help of AI become the property of a certain person (developer or user of AI).

4. An option in which a work created with the help of AI will be considered as an official work.

5. A variant in which a work created with the help of AI goes into the public domain"^{[36; p.393]}. The first two options are rejected by the author himself due to the lack of legal personality in AI. If we consider the fourth option by full analogy with the official work, then in this case the author's right should be reserved for AI, which is not possible according to the arguments outlined in our study earlier. Automatic transfer to the public domain deprives the exclusive right of the property component as such. In this regard, the third option remains the most demanding.

Speaking about inventions in general, it should be borne in mind that patent law, in comparison with copyright, is often characterized by a discrepancy between the author and the copyright holder (patent holder), as well as the fact that the assignment of exclusive rights to the copyright holder is not complicated by the transfer of these rights: the ownership of various works created by one AI to different copyright holders should be allowed if the rights to AI was legitimately transferred. If we talk about the admissibility of recognizing the exclusive rights to works created by AI for the AI system (for example, by analogy with a legal entity), then we should proceed from the definition of the exclusive right given in paragraph 1 of Article 1229 of the Civil Code of the Russian Federation. According to the said norm, it implies the use of the result of intellectual activity at its discretion in any way that does not contradict the law. It also involves extracting benefits. Based on the semantic meaning of discretion as an action of one's own volition, then AI does not have it, as well as the need for obtaining the benefits of AI is doubtful.

V.O. Kalyatin's report indicates the possibility of a number of problems in the case of securing exclusive rights to the results of intellectual activity created by artificial intelligence for the rightholder of the program that automatically generated this result of activity (a corresponding bill was submitted to the State Duma of the Russian Federation in October 2020)^{[37; p.7]}. According to the author, this approach is easy to implement, but it can lead to a monopoly of the artificial intelligence developer on the results created by the program. In addition, the use of artificial intelligence systems in the inventive process seems to him to be a significant problem, due to the growth of inventions and patents in a certain area, the emergence of "patent thickets", the associated increase in the cost of examining all patents and obtaining licenses ^{[37; p.17]}. If we consider the situation from the opposite, then if the developer is not initially granted exclusive rights, then we can predict both a decrease in the research interest itself and the interest of the specialist in property benefits, which in general may lead not to monopoly, but to a reduction in innovation processes. The arguments about "patent thickets" seem unconvincing, they are likely with the enormous growth of "strong" AI systems capable of creating complex models and designs, provided that the workload of Rospatent employees increases. However, the period of post-non-classical science as a whole is characterized by the modernization of types of work and professions, such a trend should not be an obstacle to technological changes and related revision of legislation.

V.O. Kalyatin also notes an interesting number of possible legal novels^{[37; p.21]}. Among them:

1) Making a distinction in legislation between the results of intellectual activity created directly by AI and with the help of AI. The exclusion of the human factor seems to the author to be the basis for the emergence of a special legal regime, the complete or partial exclusion of these objects from protection.

In this case, the author points out the need to protect the developer of the program, but the ways and mechanisms of this protection are unclear.

2) A novel concerning the assignment of rights to a product created by AI to a person "organizing the use of AI".

The organizer is delegated, first of all, the definition of the task assigned to the AI, the implementation of adjustments. In this regard, as it is obvious, the organizer represents a collective image of various kinds of technical specialists and users, among whom the exclusive right is divided.

3) The establishment in the legislation of a special exception in the field of copyright in relation to the possibility of using the results of intellectual activity, the exclusive rights to which belong to other persons, for the creation and training of AI.

This novel contains an element of novelty, is aimed at stimulating innovation, thanks to its implementation, a huge increase in technical solutions is possible, however, as in the case of the first novel, firstly, the legal mechanisms for its implementation are unclear, and secondly, such a presumption of unlimited use of exclusive rights for AI can give rise to new legal problems and will require serious reform of the legislation as a whole. However, it should be noted that with this approach, the rights to the result of AI can theoretically arise from the owner of the thing used to create a new product. It seems appropriate to us that this should be the right to receive remuneration as an incentive for participation with the mandatory conclusion of a written contract.

Based on the above: there are no prospects for recognizing copyrights to any object created by AI exclusively for AI, without specifying its developer, information provider, the person who provided data for the formation of machine learning, the operator who controls the operation of algorithms and is responsible for the absence of failures. Therefore, the following conclusions seem appropriate to us.

Autonomous development of intelligent solutions and their implementation by AI is possible only if there are algorithms written by a programmer. In addition to creating them, his duties include checking and correcting software, analytics, troubleshooting, etc. Therefore, anyone who makes even a relatively small creative contribution to the creation and work of AI should actually have the right to be considered the author of the result of intellectual activity created by AI. In addition, it is not unreasonable to indicate when creating an AI product that it was created by AI with a list of its authors, each of whom acquires a set of personal non–property rights: the right to be recognized as an author, the right to inviolability, protection from distortion, etc. The criterion for determining the author should be the minimum creative contribution of each of the persons responsible for all AI factors (scientific, technical, linguistic, control, etc.), on the basis of which he (AI) carries out his activities. In fact, with this approach, AI is a mediator–converter, however, due to the complex structure of the AI system, including self–learning, it cannot be said that AI is a simple labor synthesis of co-authors or a complex work. Therefore, in this case, the indication of authorship, without reference to the creation of an AI product, seems incorrect from the point of view of the nature of the origin of the result.

From our point of view, initially the exclusive rights to the AI program code should be held by a programmer who can transfer them, including through a contract. In fact, if the AI was not transferred, was not used and was with the developer of the program, it does not mean that there were no exclusive rights to it. In addition, according to the current legislation, the author owns authorship and other personal non-property rights (paragraph 2 of Article 1228 of the Civil Code of the Russian Federation), he also initially has an exclusive right to the created result of intellectual activity (paragraph 3 of Article 1228 of the Civil Code of the Russian Federation). Based on this, the exclusive right to the products (results) of intellectual activity created by AI should initially focus on the authorship of an invention, but taking into account its regime (for example, an official work), as well as the transfer of these exclusive rights. Thus, the exclusive right may belong not only to the AI programmer author, but also to the secondary copyright holder of the program used to create the result of AI intellectual activity.