Protection of intellectual property rights in artificial intelligence
Dušan Kovačević is a lawyer who holds a master’s degree on the topic “Protection of intellectual property rights in artificial intelligence” from the University of Tartu. As he emphasizes, this Master’s program has provided him with a deeper understanding of the modern world and the factors that influence it.
“I learned a lot from the master’s degree, which completely changed my perception as a lawyer, but also as a citizen, in terms of understanding the interaction of both rights and obligations, and civil liberties through information technologies, because at that level today’s world functions,” explains Kovačević at the beginning of the interview for WebMind.
However, the work he wrote with his colleague Olga Fesenko refers to artificial intelligence in the plural, which can seem unusual, but is justified.
“As I understand it, there is initial code, training, use and economic exploitation. Computer code as such is more or less protected as a literary work, in that aspect, there is a person who writes the code, who would be the author of that work, then a person who trains the code, who could be covered under the rights of the interpreter. Anything after that, the use would be in the form of economic exploitation, which is a transferable right, in accordance with time and territorial constraints. What AI creates will relate precisely to who has the right to economic exploitation. We can think of AI as an incredibly advanced randomizer, but it’s so advanced that it gives you random output within the expected limits,” he explained.
“AI can’t be an author, just a tool”
Although the work was created back in 2017, it can still be considered current in the context of understanding intellectual property, especially because it envisaged the establishment of what we now know as the AI Act. In addition, the paper points to one paradox – most international treaties concerning copyright were passed before the expansion of artificial intelligence and therefore didn’t precisely define that the author must be human.
“It was never considered that anyone other than a human being could purposefully create a work of authorship. The entirety of intellectual property rights was created by certain philosophical schools that believed that the author’s work arose by importing the creation into a particular work, giving it meaning, and a concrete intention to create the work. In this paper we focused on the analysis of philosophical thinking as an understanding of the author’s work of man’s creative power. It was evident that technology itself, AI itself, cannot be an author, in principle it can only be a tool.”
For a better understanding, we asked the interlocutor to explain the concept of authorship in artificial intelligence in the gaming industry, i.e. on the example of creating a video game.
“The tool itself and the legal regulation of the use or training of that tool will dictate ownership of the very result that ejects AI. If we have a video game that has multiple elements, we could find specific people who would claim certain aspects of it and who could be listed as co-authors. If the game involves alternations that come up with a separate product, it depends on how it is regulated within the accompanying documentation, what we call Terms and Conditions. The most famous example that I remember now is, after Dota started its own life in Dota 2, Blizzard was butthurt and then, when Hammer Reforged was released, the Terms and Conditions stated that everything that is created within the ecosystem of the mapmaker itself or glitches, alterations, even usernames remain their intellectual property.
“Indifference to other people’s rights and obligations”
If we recall recent events in the entertainment industry, such as creating music using AI, which is trained on copyrighted songs, the problem of interaction between the data that AI is fed with and the copyright that people claim on it is obvious.
“People missed that step because tech people like developers don’t practice law until the sky is falling. It should be taken into account that any information covered by intellectual property law or some of the related rights should not be included in this training dataset. Developers do not think about it initially because they are more focused on creating products and results, and here we see indifference to other people’s rights and personal obligations,” says the interlocutor.
This is supported by the statement of the CEO of the platform Midjourney, who said that he was aware that they were not entitled to the visual material they used during the training, but that, in his opinion, it is impossible to trace the ownership of more than 10,000 photos.
“Specifically in the case of Midjourney, people have the ability to see a certain thing and create new content in relation to it, in terms of understanding the essence of the concrete thing they are looking at. AI can only make an amalgam of what’s in their training dataset and nothing beyond that. They can improve themselves by including their own creations in the training dataset, along with new inputs. You can never produce something that wasn’t covered during training. On the other hand, people have the ability to create things regardless of whether they have seen something or not,” Dušan pointed out.
Why is artificial intelligence a challenge to intellectual property?
Bearing in mind the above problems, the so-called “legal limbo”, we asked the interlocutor whether the anger of people who feel threatened or deprived of rights is justified.
“I have a professional deformation related to the fact that the whole life is regulated by the law, whether we like it or not, to what is the permissible level of alpha toxins in milk. Anger can be justified or unjustified in the context of how the rules are understood. At the legal level, whether that person claims copyright to the photo itself, if the conditions of economic exploitation are met, I don’t see the difference between using AI tools and Photoshop. We are entering a discussion related to changing the whole civilization because it is not only the creative industries that are in trouble, there are lawyers and journalists and so on. One colleague of mine wrote, “Ten years from now, we won’t be lawyers or programmers; We’re going to be prompt engineers.” And your only task will be how to talk in a special way with AI to get the desired result, within what you need for your profession. We can be angry or not, but the fact is that all the countries in the world can’t stop it. The only thing we can do is adapt to the new economic reality that in 10-20 years from now, I believe there will be technical implementations that will create content on a personal level, for example, you will have a TV show tailored for you or stories for children that on demand,” the lawyer says, adding that for him a much more interesting question is at what point will AI reach the legal standard where we can equate it with the human one.
“I don’t have an answer to that, but I think it should be part of the discourse because the moment it becomes relevant, it will be too late,” he concluded.