By: Luca Schirru [1] and Allan Rocha de Souza [2]
On December 11, 2024, the Brazilian Association of Intellectual Property (ABPI) organized and hosted the Forum “Decoding Copyright in the Age of Artificial Intelligence” in São Paulo. The event was on December 11th, just one day after the Brazilian Senate approved the AI Bill (Bill No. 2338/23). The bill will now be returned to the House of Representatives for further discussion and, if approved, will be submitted for presidential sanction.
The forum brought together a diverse group of stakeholders: IP rightsholders, technology companies, researchers, and legal practitioners to explore some of the most pressing issues at the intersection of copyright and artificial intelligence (AI). Discussions were particularly focused on the complex challenges posed by generative AI in the realm of copyright. The adequacy – or not – of the AI Bill in addressing issues such as remuneration, opt-out mechanisms, transparency, and text and data mining was a central topic of discussion, featuring insights from national and international speakers.
The panels covered issues regarding AI inputs, outputs, and the balance between copyright protection and the need for innovation and development. The Brazilian Copyright Institute was represented by Allan Rocha and Luca Schirru.
Bertin Martens offered an economic perspective and reasons why opt-out mechanisms and restrictions on AI training may be economically inefficient, also referring to the provision in the Brazilian AI Bill. Luca Schirru addressed the challenges related to training AI systems and text and data mining. He highlighted how the current text of the Bill fails to adequately promote research due to several shortcomings, such as (i) the lawful access requirement and (ii) the fact that the copyright exception is restricted to institutions and (iii) excludes university-companies partnerships.
Caroline Tauk addressed key copyright challenges related to AI-generated outputs, such as authorship, ownership, originality, and potential copyright infringement claims. Professor Pamela Samuelson provided an overview of the legal challenges and their competing arguments presented in U.S. court cases. Samuelson’s insights spanned both input-related issues, such as data used for training AI systems, and output-related concerns, highlighting the complexity of navigating copyright in the context of AI.
The approval of the AI Bill faced criticism since the final version of the text was made available less than a week before the vote, as noted by Andriei Gutierrez, who considered it to be “irresponsible that the bill passed in the Senate. The text was approved last Thursday by a minimum number of senators, and the rest had until Tuesday to approve it.” (see here and here) Finally, Allan Rocha de Souza presented a critical perspective on the AI Bill and explained why, as it is, it does not strike a good balance between the promotion of research, authors’ remuneration, or an innovation environment, and suggested that it could be greatly improved with minor changes.
The legislative year will resume in March 2025 and given the broad criticism towards the approved AI Bill from various stakeholders, further discussions and revisions to the copyright provisions are expected in the House of Representatives.
[1] Executive Director and Researcher at the Brazilian Copyright Institute. Postdoctoral researcher at INCC. Copyright Professor at the Specialization Program on Intellectual Property Law at PUC-RJ. Lawyer. Contact: luca.schirru@ibdautoral.org.br – ORCID: 0000-0002-4706-3776.
[2] Copyright Professor at the Graduation Program on Public Policy, Strategies and Development (PPED/UFRJ) and the Civil Law and Humanities Department of the Federal University of Rio de Janeiro (DDHL/ITR/UFRRJ), and in the IP Specialization Course at Pontifícia Universidade Católica (PUC-RJ), Brazil. Scientific Director of the Brazilian Copyright Institute (IBDautoral). Copyright consultant at Fundação Oswaldo Cruz (FIOCRUZ). Lawyer. Contact: allan@rochadesouza.com – ORCID: 0000-0002-6549-0085.
