Artificial intelligence as “author” and “user” of works and other objects of protection

Artificial intelligence (AI) is already in our everyday life, it is used to create code, computer games, audio-visual works and journalistic materials. Generative AI is not just a tool that supports the creative process, it makes decisions without human intervention. This development comes with its legal consequences – including from the point of view of intellectual property rights.

A few days ago, in “Economic” magazine, Atty. Emiliyan Arnaudov examined the main points related to the interaction between artificial intelligence and the regulation of copyright and related rights. This was his next participation on this current topic after he was a panelist in the event “Artificial Intelligence and Cyber Security”, held by the Confederation of Employers and Industrialists in Bulgaria and later he gave interviews on the topic to the Bulgarian National Radio and Bulgaria ON AIR television.

This interaction is two-way, since AI is the user of numerous objects protected by copyright and related rights, and on the other hand, AI produces its own product, for which the question is raised whether it is subject to such protection, whether it has copyright. Of these two aspects, the regulation of the first is significantly more advanced, as in 2019 the EU adopted the Directive on Copyright in the Digital Single Market.

Atty. Arnaudov led the team that drafted the changes to the Copyright and Related Rights Act, dictated by the new European Directive (EU) 2019/790 on Copyright and Related Rights in the Digital Single Market. This directive is to be introduced as part of the national legislation and in mid-April 2023 the bill was submitted to the National Assembly. The texts related to the regulation of the use of protected objects by artificial intelligence are provided in the Copyright and Related Rights Act.

Artificial intelligence, on the one hand, is a user of protected content, and the Directive makes it easier in certain cases for it to use it without prior permission and payment in order to stimulate scientific progress. On the other hand, it creates new texts, images, music, and the problem arises whether someone has copyright and related rights on what it created, and if so – then who? The conflict comes because it uses the achievements of the creative sector for free and thus generates content that could compete with man-made content and deprive the creators whose works he used as a base of future income.

Exactly because of this some of the most debated questions related to the regulation of the rights to the content created by artificial intelligence are “Can the result of the activity of artificial intelligence enjoy copyright protection, since it is not the product of human authorial activity? If it can, who owns the copyright?” In the US, the Supreme Court has ruled that AI cannot own patents. The Copyright Office states that a work created without human intervention cannot be subject to copyright. Outside the US and Europe, China has been integrating the use of artificial intelligence for the past year to improve its judicial system. With it, laws are made, documents are drawn up, charges are brought, “human errors” are looked for in decisions. Judges must consult the AI. Although there is a need for comprehensive regulation of artificial intelligence, for now the new changes address how AI uses copyrighted and related rights objects, but not how rights are settled if the product is created by the artificial intelligence itself.

In order to use an object protected as intellectual property, in principle, prior permission must be obtained. It is usually associated with a fee for use. In the Directive, two important exceptions are provided for processing and automated analysis of text and data (data mining) from the general rule requiring the consent of the right holder to use the protected object. The directive says that if the right holder wants to prohibit the use, and the object is available on the Internet, this prohibition must be accompanied by automatically recognizable means that mark it as prohibited, i.e. the prohibition must be indicated by something that can be automatically recognized by the software that views it, something incorporated inside the code of the page as an appropriate indication to the software that this part cannot be used in automated processing.

However, when the right holder has not expressly prohibited the AI from using the protected object with a special code, then with automated processing of text and data, this object can be freely used. But there is also a wider possibility that applies only to processing for scientific purposes – then even if the right holder has prohibited it, the object can be used for automated processing. Such an exception will be made for such types of scientific research organizations that are entirely for public benefit such as the Bulgarian Academy of Sciences, universities, cultural heritage preservation funds, etc. AI cannot function without using large amounts of data. For example, if certain literary works are being investigated, it could go through the analysis of hundreds of thousands of texts to arrive at the result of the relevant scientific research. Therefore, the EU believes that technical progress and the development of new technologies should not be hindered by the instruments of copyright and related rights. The specified exceptions regulate only this first aspect, in which the AI is a user of objects protected by copyright and related copyright.

Far more debatable is the second aspect, where AI produces certain objects that are disputed as to whether they should be protected as the object of copyright and related rights. If they are protected, who would be the right holder – should it be the developer of the AI, its owner, user, or the assignee of the task. In the US, the Supreme Court refused to hear an appeal by a scientist who sought patents for inventions created by his artificial intelligence system. The US Copyright Office in March 2023 issued an opinion that if the creation is entirely the work of AI without any human intervention, it has no copyright protection. But if a person arranges a comic of images created by AI, then this comic is subject to protection, the so-called collective work. Bulgaria, and the EU as a whole, are far from resolving this issue related to the copyright of AI creations. Legislations will still seek a balance. There are various options, and even some innovative ones can be offered. For example, the creation of an AI can be considered as a folklore work, since the final result is obtained after the works of many persons of different generations have been processed, and there are related rights of the producer or publisher and possible performer on it. There is no copyright in folklore works, only related rights are protected, which are granted to an organization or person in the role of publisher or producer who invested in the creation of the object in question. Another possible solution is to provide for the content created by artificial intelligence to be mandatorily managed by collective rights management organizations, thus providing those engaged in the creative sector with an income compensating for their losses and rewarding the efforts of the multitude of creators whose objects have become base for what was created by artificial intelligence. In this way, a compensatory mechanism can be created for them all to receive an income after being given free access to their creativity.

As much of the AI-generated content competes with human-generated content, the Fourth Robotics Law, introduced by Lyuben Dilov – father in 1974 is likely to be enshrined at the European level, which states that a robot is obligated under all circumstances, to legitimize itself as a robot.

Hopefully, the regulation will come into force soon, providing the necessary answers to some of the important legal questions that AI poses.

Photo: Bulgarian National Radio