Brazil Urges Stronger Enforcement Mechanisms for AI and IP Rights
Brazil’s statement at WIPO CDIP 33 emphasized AI’s transformative potential in IP while highlighting challenges like copyright violations and the need for fair remuneration. Advocating for global governance and legislative reforms, Brazil called on WIPO to ensure transparency, equity, and inclusive policies that protect creators’ rights and provide equitable access to AI benefits. Brazil CDIP 33rd Statements BRAZIL: Thank you, Madam Chair. We thank WIPO for the opportunity to discuss AI in the context of development. We note the many number of WIPO representatives at the podium now including ADG Natsumi which we thank very much for a time showing that this is a cross organizational aspect and expresses the correctness of treating this topic at the CDIP without prejudice to the specific technical committees of WIPO. Now, Madam Chair, the rapid pace of development of AI technology has put it under the spot. In 2022, sorry, we were all stunned by the launch of generative AI systems which we all assumed were years away and now we as countries and we as international organizations are analyzing its implications, potential uses and the necessity of regulating it. AI, of course, can be of significant assistance to IP offices in the process of registration and management of industrial property rights such as patents, trademarks, industrial designs. The translation of technical documents is another area where AI can be widely applied with the assistance of WIPO. Some such uses were explored recently in the last session of the SCP and application of AI in PCT activities are also being discussed at the PCT Working Group. But another area of analysis is the constitution and exercise of Intellectual Property Rights. So on the patent side, the use of AI systems can have an impact on new discoveries, increasing innovation. It also raises questions about patentability and ownership of the innovation resulting from such systems. This was explored previously by the Secretariat. So AI can assist research with the potential to result in a large number of patent applications. Additionally, it may be used to boost anti-competitive strategies adopted by specific industries to file patents for incremental modifications aiming to block competitors’ R&D, so-called patent tickets. So in this context, legislative changes in the patent legislation may be necessary to encompass and address such challenges and opportunities. But the speed with which this technology develops is, of course, a challenge for us as regulators. Now, I would like to address specific copyright-related issues on AI. As we all know, AI systems, especially generative AI, demand vast amounts of data, the most valuable of which is copyright protected. Copyrighted works such as music, books and movies contain organized expression of ideas which are vital for AI systems to develop their capacities to, so to speak, mimic human behavior. Of course, generative AI can be a tool for creation and like any new tool, it has benefits and risks. It raises a series of ethical, aesthetic, technical and legal questions that we have the responsibility of examining. Now, in many countries, including Brazil, the United States and others, the protection of an author’s right is a constitutional clause. Article 27 of the Universal Declaration of Human Rights protects the moral and material interests resulting from any scientific, literary or artistic production of which the person is the author. So we see a strong basis for discussing the protection of author’s rights in AI systems. There’s a trend towards the reduction of the value of creative works, the so-called commoditization of creative works, which affect the enjoyment and exercise of the copyright protection to authors, artists and creators. A study by CISAC published just last week shows that copyright right holders will lose 22 billion euros until 2020 in the music and audiovisual sectors alone. This figure does not include losses to writers, photographers, interpreters or producers of phonograms. Another study by German and French collective management organizations estimated that music authors will lose 27% of the royalties due to generative AI by 2028, four years away. Similar negative effects have been estimated in Australia and New Zealand. Now, what are the causes of such losses? They will be due to the substitution of human labor for AI, the competition of AI produced content with copyrighted works and the lack of payment for the use of protected works in the training of AI systems. This raises a particularly important question relating to the long-term production of works. AI systems demand a diversity of cultural production, but how could we generate it without authors, performers, creators and humans? How can we continue to have the production of works if we don’t have the necessary economic incentives for human authors in the copyright system? Further, how do we differentiate authentic human work from AI-generated works? Considering this scenario, it is without a surprise that authors demand a fair remuneration for the use of the works. Many letters, public manifestations have proliferated with thousands and thousands of artists, performers and their associations signing. At the same time, tech companies argue this would stem innovation, but the individual market value of a leading AI company can reach 250 billion U.S. dollars. Reports in the media also indicate that companies such as Google and Microsoft spend every few months 10 billion U.S. dollars on data centers and AI infrastructure. Further, as I had mentioned, copyrighted works are one of the most important, if not the most important input for generative AI systems. It would be not only illegal but absurd if AI companies did not pay for other basic inputs such as energy, data centers, or software. Why would this be the case with copyrighted works? And according to reports by authors, artists and even representatives from AI companies in the media, use of protected works is happening without authorization from rights holders. As a consequence of this, companies developing AI systems may have violated in each work used at least five rights granted to authors by copyright legislation. The first
