The development of artificial intelligence (AI) raises the question of who should own the rights to the objects it creates. Lawyer Anton Pchelkin discusses the results of AI activity from the point of view of copyright
Why is this important?
Ever since the Middle Ages, people have dreamed of creating a mechanical, artificial likeness of life. Now man seeks to develop artificial intelligence. And although the level of technology does not yet allow the creation of a fully functional artificial intelligence capable of self-awareness and thinking, advances in mathematical modeling and neurophysiology allow the development of self-learning systems capable of performing simple tasks with very high accuracy.
AI has proven to be more than just a predictive or data analysis tool. Today, such systems are able to create literary, musical and visual works, largely imitating their original authors. This imitation determines the content of the current stage of development of AI systems, since human-made materials are used for their training.
In modern systems built on the basis of AI, there is a very high creative potential that can, if not replace human authors, then certainly make up a large share of the production of entertainment content. Sooner or later, humanity will think about whether it is possible for AI systems to have any rights, including the objects they create.
In the legislation of a number of countries, at the end of the 20th century, norms appeared that regulate the emergence of rights to objects of intellectual activity created with the help of a computer. They are the result of a long collaboration between legislators, IT experts and the business community.
As for domestic legislation, within the framework of the federal project “Regulatory regulation of the digital environment”, it is planned to regulate legal relations in the field of robotics and the use of AI technologies. In particular, it is planned to approach the issue of the mode of objects created using AI systems.
At the same time, for many specialists in the field of law, the resolution of this issue is seen only in conjunction with the solution of a larger problem related to the consideration of AI systems as subjects of law.
For whom is it important?
The importance of the issue of the legal regime of objects created with the help of AI is growing along with the development of the entertainment industry.
Today, we hear more and more about the use of AI in the creation or processing of videos, photos, music and texts. And there is an increasing likelihood of practical problems: from disputes over authorship to claims of non-protection of works created with the help of AI.
Perhaps the level of technical development is not yet too high for the emergence of such problems right now. But the very existence of such a perspective requires us to understand the subject area of both copyright and the functioning of AI systems.
Creative AI in the creative industry
In relation to our topic, it would be most correct to classify AI systems by creativity. This potential is determined by the algorithms and sets of rules by which the system operates. The will of the creator of the system is also present here: the higher the level of freedom of the machine, the less predictable the final result. Classifying AI objects according to the degree of impact on the final result, you understand that only a small part will fall into the category corresponding to the full cycle of creative work, where the final result is almost unpredictable.
Three such categories can be distinguished:
- Little effect on the final result. May have an indirect effect on the result. AIs belonging to this category do not show “machine creativity”. This category includes AI used in text editors to correct errors and styling, in cameras to intelligently adjust image settings, and in audio editors for noise reduction.
- Affects the final result, but is used as a tool as intended by the author (human). To some extent, there are signs of machine creativity. Freedom is limited by context or a small amount of data for analysis and learning. We are talking about AI used to change images in graphic editors, to restore image fragments.
- The final result is unpredictable. Creating the final result is the purpose of the functioning of AI. Such AI objects include self-learning systems. A striking example are systems using the competitive principle (GAN) and evolutionary algorithms.
Naturally, systems from the last group are of the greatest interest from the point of view of copyright, since they are able to function, although according to the author’s intention, but their results practically do not depend on human actions.
Human role
Any, even the most perfect AI, has a creator. If the final result of the work of AI is unpredictable and original, the conditions for its functioning were still created by a person: be it programming, collecting and systematizing data for system training, technical support. We must not forget about the Garbage in – garbage out principle, according to which, when training AI on low-quality data, the result will also not be of high quality.
Human labor occupies a central place in the copyright law of the countries of the common law system (common law, Anglo-American case system. -). In resolving disputes about authorship, until recently, the sweat of the brow doctrine was applied, which suggests that in the absence of originality, but the significance of the effort expended, the creator of the work has the right to protection.
Proceeding from this concept, in the legislation of a number of countries (Great Britain, New Zealand) there have long been rules that extend copyright protection to works created by a computer. The author is the person who created the necessary conditions for the creation of the work. It can be either a programmer or a user.
What and who can have rights?
According to Russian intellectual property law, authorship and creativity are presumed (implied by default. -). The first is by virtue of Article 1257 of the Civil Code of the Russian Federation, according to which the person indicated on the original or copy of such a work is recognized as the author of a work. The second is due to the indication that the lack of novelty, uniqueness and originality of the result of the work cannot indicate that such a result was not created by creative work.
In addition, creativity does not exist without the figure of the author – the entire corpus of copyright revolves around the subject – an individual – and the rights arising from his activity. This could be contrasted with the idea that AI systems have legal personality—that is, their ability to acquire rights and obligations through their actions. But such a concept is still only at the stage of discussion and is unlikely to be reflected in legislation in the near future.
From this we can conclude that in the normative plane not only there are no criteria for differentiating “human” and machine creativity, but the latter as a whole is not possible.
The objects of copyright are works of science, literature and art, regardless of their merits, purpose, and mode of expression. Even in the absence of the ability to recognize the AI system as the author, the images, texts, and musical works created by him meet the established requirements.
Returning to article 1257 of the Civil Code of the Russian Federation, the person by whose will the AI created the work can easily indicate his own authorship. In this case, the author of the work will have all the rights provided by law. It appears that naming a programmer or user of an AI system as the author of a work would be in line with the common law approach.
Using a more formal approach, we can talk about the presence of a work in the absence of the author as such. It seems that such works may pass into the public domain upon their creation – since the author is absent. At the same time, the person who directly organized the creation of such a work by artificial intelligence has the opportunity to publish it, and in connection with such disclosure, related rights may arise – the rights of the publisher, including the exclusive right to the published work and the right to indicate his name on copies of the published work. .
Publisher rights are similar to copyright, but much narrower. Unlike copyright, related rights will be valid for 25 years, not 70. However, in order to achieve an economic effect – making a profit and returning the funds invested in the creation of an artificial intelligence system – such a period will be sufficient.
Conclusions
Rejecting the possibility of qualifying the activity of AI as creative, but not fixing the order of emergence and distribution of rights to the obviously existing creative results of such activity, we may encounter the difficulty of fully introducing the results of AI activity into civil circulation.
So far, in case of disputes regarding the distribution of rights to works created by AI, there are two options: indicate the programmer or user as the author, or recognize the work as not protected by copyright and carry out its lawful publication, which leads to the emergence of related rights of the publisher .