An agreement signed between the government and the opposition allowed the Senate Special Committee to approve, by symbolic vote, the bill that regulates Artificial Intelligence (AI) in Brazil, this Thursday (5). That’s because there is a dispute around of the sections that require AI systems to respect the integrity of information and combat disinformation.
The Senate Plenary approved, also this Thursday (5), the urgency of the text and the expectation is that the merits of the measure will be analyzed in the House Plenary next Tuesday (10), when the parties will still be able to suggest new changes to the text.
To be able to approve it by symbolic vote – when there is an agreement and parliamentarians do not register their vote in the system –, rapporteur Eduardo Gomes (PL-TO) removed some excerpts that mentioned the integrity of the information. On the other hand, Gomes maintained the provision that says that the integrity of information is one of the principles that govern the use and development of artificial intelligence in Brazil.
The government leader in Congress, senator Randolfe Rodrigues (PT-AP), assessed that the agreement did not harm the principle that AI systems will have to respect the integrity of information.
“There was the exclusion of a somewhat confusing device, which the rapporteur accepted to build this unit, but it does not harm the principle of information integrity”, explained Randolfe.
The opposition leader, senator Marcos Rogério (PL-RO), maintained that his concern was that these excerpts would harm freedom of expression.
“From the point of view of what I defended, we improved the text. Points that still need care here or there still exist, but the text will still pass through the Senate Plenary and the Chamber of Deputies, and then return to the Senate if there is a change in the Chamber. I am happy with the work of the rapporteur who, in an environment of great divergence, built a path of convergence”, explained Rogério.
The director of the Recife Institute of Research in Law and Technology, André Fernandes, who is part of the Coalition Rights on the Network, highlighted the Brazil Agency that changes regarding the integrity of the information dehydrate the text, but may have limited effect.
“It was the agreement for the text to move forward, within the logic that the rapporteur’s base, who is a Bolsonarist, was pushing for the removal of these elements. And that’s a problem. Obviously, it is yet another dehydration of the text that may be harmless in the end because the integrity of the information is still the basis [da IA no texto]”, he explained.
Changes
Rapporteur Eduardo Gomes (PL-TO) excluded from the text, at the request of the opposition, two provisions that cited the need to observe the integrity of information in AI systems.
In the previous document, it was foreseen that, to classify whether or not an AI system is high risk, it was necessary to analyze whether the technology could cause “risk to the integrity of information, freedom of expression, the democratic process and political pluralism”. This excerpt has been deleted.
The rapporteur also excluded Article 31, which provided that the developer of an AI system, before making the software in the market for commercial purposes, should “guarantee the adoption of measures to identify, analyze and mitigate reasonably foreseeable risks with regard to fundamental rights, the environment, the integrity of information, freedom of expression and access to information”.
On the other hand, the rapporteur maintained, among the foundations for the use and development of AI systems in Brazil, the following criterion: “integrity of information through the protection and promotion of reliability, precision and consistency of information to strengthen freedom of expression, access to information and other fundamental rights”.
The bill defines that information integrity is “the result of an information ecosystem that enables and makes reliable, diverse and accurate information and knowledge available in a timely manner to promote freedom of expression”.
Project
Authored by the President of the Senate, Rodrigo Pacheco (PSD/MG), the text establishes the fundamental principles for the development and use of AI. It defines that technology must be transparent, safe, reliable, ethical, free from discriminatory bias, respecting human rights and democratic values. Furthermore, the project requires that technological development, innovation, free enterprise and free competition be considered.
The text also provides for respect for the copyright of artists and journalists who have their works used by AI systems and provides for an authority to govern AI regulation in Brazil, with the creation of the National AI Regulation and Governance System (SIA) .
This System will have a residual role, with the main regulatory role remaining with the current sectoral regulatory agencies, which will still have an emphasis on the SIA.
According to the project, the SIA will be responsible for “exercising full normative, regulatory, supervisory and sanctioning powers for the development, implementation and use of artificial intelligence systems for economic activities in which there is no specific sectoral regulatory body or entity”.
The project also defines which AI systems should be considered high risk and, therefore, need stricter regulation, in addition to prohibiting the development of technologies that cause harm to health, safety or other fundamental rights.
The project prohibits, for example, the Public Authorities from creating systems that classify or rank people based on social behavior to access goods and services and public policies “in an illegitimate or disproportionate way” or that facilitate the abuse or sexual exploitation of children and teenagers.
