Juliana Sakai (Transparência Brasil) Visits Seattle University

The visit was covered by Serafin Deleon Guerrero, Naja Johnson, Ivonne Lares, Maryjose Ortega Ortiz, Carmen Ruiz-Zorrilla Garzón, Maya Walthall, and Zoe Zepeda Garcia - students in Dr. Onur Bakiner’s Comparative Politics class.

Sakai states that the laws governing AI in the nation are changing significantly, and now include a framework to address the moral, societal, and financial ramifications of AI technology. The talk highlights Brazil's efforts to improve transparency, governance, and accountability, particularly through anti-corruption initiatives, the Freedom of Information law (2011), and public data collection to strengthen civil discussions.

Of particular note are Bill 21 (2020), which proposes principles and rights but lacks practical risk management tools. The bill offered no concrete tools to address risks of the technology. Its interpretation of civil liability as subjective, which goes against much established precedent, has been criticized.

The perceived weaknesses of that law led to the drafting of Bill 2338 of 2024, which “guarantees basic rights for people affected by AI”, establishes governance for high-risk systems, and addresses structural racism in AI applications, emphasizing consumer rights and objective liability for AI operators. Its risk-based categorization of AI systems resembles the European Union’s approach in the AI Act.

In addition to bills, Sakai mentioned important lawsuits. IDEC v. Herring and IDEC v. Sao Paulo Metro both bring up situations where facial recognition data was collected without consent by private and public entities. Another case concerns Meta: when Meta decided to start using personal data of the platform’s users as training data for its generative AI, and changed its privacy policy toward that end, the National Data Protection Authority (ANDP) in Brazil decided to suspend the new privacy policy, as it affected the fundamental rights of Meta’s users. The problem according to them was the lack of information, so the solution they proposed was to send a notification about the processing of data for use in generative AI. Users now have the right to opt-out.

Perhaps the most famous case coming from Brazil is Xandão v. X. In Brazil there is a civil rights framework for the internet, in place since 2014, and the civil code obligation for foreign companies to respond to judicial orders, as well as to have a legal representative in Brazil. The X case started when X, the former Twitter, received several court orders to remove illegal content and fake news from the platform. These warnings were all ignored, which led to the situation escalating and a confrontation between Elon Musk and the judge from the Supreme Court in charge of the case (Alexandre de Morães). Musk challenged Brazilian justice, and as a result X was banned from the country until he complied with the judge’s demands: pay the fines and establish a legal representative. After four days, Musk gave in and X came back to Brazil.

To conclude: Sakai situated the legal and judicial developments around AI regulation in that country’s broader legal system of fundamental rights, and consumer rights. A comprehensive AI law is still lacking, but the existing legal norms can guide the country’s approach to AI regulation in the future. The judiciary has grown in power in the absence of specific legislation. And finally, a major gap Sakai identifies is the absence of data privacy and protection in cases involving public security.