The recent wave of artificial intelligence (AI) generative, led by ChatGPT, has brought with it some problems. Among them is the difficulty in recognizing the intellectual property rights of the works on which these tools are based to create texts, images, and other content. Therefore, legislation addressing this issue is urgently needed, according to Victor Habib, a lawyer, professor, researcher, and author on digital law.
His article “The Role of Intellectual Property in Fostering Innovation: A Legal Perspective on Creations by Artificial Intelligence”, second place in the 1st Ericsson Award for Academic Production on Intellectual Property, investigates the topic, which guides a chapter of a book on digital law and innovation that is about to be released.
"We don't know where the data feeding these AIs comes from. For example, AI software that generates images needs to be based on other images—and the artists who created those images have copyright. Because of the lack of regulation in this area, which is crucial, we don't have transparency regarding what was used to generate a particular image," he explains.
Another obstacle is that AI can become almost like a black box. Often, not even the companies that own these tools can explain how they arrived at a particular end result. Thorough investigation is required to understand the logic behind their creation. Some AIs, such as ChatGPT on Bing, specify the sources on which it was based, as a way of getting around this problem, which can be a little more complex in some situations.
Regulation
For Habib, regulation is important so that artists, intellectuals, and others can have their copyrights protected. Furthermore, the legislation will help protect the ownership of AI itself for creators and the companies that invested money in its development. He advocates for theorists and thinkers from the legal and technological fields to participate in the discussion.
This is a unique case—there is no case law on the subject. Intellectual property law states that to hold copyright on something, the author must have thought, planned, and executed the creation of a given work. In the case of a work created by AI, the AI creator has influence, in the sense of providing data, but not in the final result. They are not the ones who think, plan, or implement it. There are several types of generative AI, and each one works differently.
"There's no law at the moment. It's a no-man's land these days," he stated. "The sooner we have regulations, the establishment of clear rules for innovation, in terms of intellectual property, the better. You know the rules of the game. There's legal certainty, and it's better economically for the country and the entire surrounding ecosystem," he argues.
DABUS
In his article, he discussed the case of DABUS (Device for the Autonomous Bootstrapping of Unified Sentience), an AI system that created two inventions and was conceived by Stephen Thaler. The engineer filed lawsuits between 2018 and 2019 in Europe, the United Kingdom, the United States, Australia, New Zealand, and South Africa, requesting copyright recognition of the AI. In this case, DABUS would be considered the inventor, and therefore the intellectual property would be his.
The response he received in almost every country was that copyrighting AI was not possible. The only place Thaler succeeded was in South Africa, which recognized DABUS's intellectual property. However, the country's judges failed to address other issues arising from this decision. Where does the money generated from royalties go, for example? To the AI or its creator? The debate could still go to the final court in the African country.
"All decisions haven't focused on the degree of AI accuracy, what constitutes originality, what constitutes creativity. Because there's no legislation, no one has any basis for making decisions about this. We depend on legislators to provide some basis for us to move forward, because otherwise, this chaos will continue, like people selling books made by ChatGPT on Amazon," he states.
Full report below:
Português do Brasil 
English