Support Centre

You have out of 5 free articles left for the month

Signup for a trial to access unlimited content.

Start Trial

Continue reading on DataGuidance with:

Free Member

Limited Articles

Create an account to continue accessing select articles, resources, and guidance notes.

Free Trial

Unlimited Access

Start your free trial to access unlimited articles, resources, guidance notes, and workspaces.

Brazil: Senate considers bill on regulating AI

This story has been updated - please see the most recent update below.

The Brazilian Senate announced, on 12 May 2023, that it will analyse Bill No. 2338 of 2023 to regulate artificial intelligence ('AI') systems in Brazil. In particular, the Senate highlighted that the bill creates rules for making intelligence systems available in Brazil, establishes rights for people affected by their operation, and provides for penalties for violations, as well as information regarding the supervising body.

More specifically, the bill specifies requirements for the operation of AI systems in Brazil, including a requirement that such systems undergo a preliminary assessment carried out by the suppliers themselves, in order to determine whether they can be classified as being of high or excessive risk. To this end, the bill outlines systems that will be considered high risk, namely those that will be used for activities, such as:

  • credit rating;
  • identification of persons;
  • administration of justice;
  • implementation of autonomous vehicles;
  • medical diagnoses and procedures;
  • decision-making about access to employment, education, or essential public and private services;
  • evaluation of students and workers;
  • management of critical infrastructure, such as traffic control and water and electricity supply networks; and
  • individual assessment of the risk of committing crimes and personality traits and criminal behaviour.

On the above, the bill provides specific requirements for high-risk systems, including a mandatory algorithmic impact assessment. More generally, the bill also provides restrictions on how AI can be used, noting that the same cannot be used for subliminal techniques to induce people to behave in a way that is harmful or dangerous to their own health and safety; nor may it exploit specific group vulnerabilities, such as those associated with age or disability, to induce harmful behaviour.

In regard to individual rights, the bill highlights that individuals must be provided with the right to:

  • contest and request explanations on decisions made by these systems;
  • request human participation in the decisions of these systems in certain situations;
  • obtain information about its functioning; and
  • not to be discriminated against and request correction of discriminatory bias.

Furthermore, the bill stipulates that AI systems must adhere to transparency requirements, namely whenever such systems are used, individuals must be informed.

You can read the bill here and track its progress here, both only available in Portuguese.

UPDATE (29 November 2024)

Commission publishes report on new amendments to AI bill

The Temporary Commission for Artificial Intelligence in Brazil released, on 28 November 2024, its report analyzing proposed amendments to the bill received since the Commission's previous report.

The approved amendments included the following changes:

  • to limit the scope of the bill to exclude the use of artificial intelligence (AI) by individuals for non-commercial activities, as well as to exclude from the scope of the bill scientific, testing, and system-developing activities, in order to stimulate innovation;
  • to exclude 'large-scale' from the criteria that define high-risk AI;
  • to reinforce the protection of children and adolescents in the context of AI use, by strengthening the language concerning vulnerable groups;
  • to improve the role of sectoral authorities in the notification of incidents and registration of high-risk AI systems; and
  • the introduction of specific obligations related to general-purpose AI based on 'systemic risk,' defined as potential negative adverse effects arising from general-purpose, generative AI systems with a significant impact on individual and social fundamental rights.

You can read the report and the latest version of the bill here and track its progress here, all only available in Portuguese.

UPDATE (5 December 2024)

Commission approves AI bill

The Temporary Commission for Artificial Intelligence approved on 5 December 2024 the report on amendments to the bill. The Commission highlighted that the report includes provisions provided under Bill No. 21/2020 that establish foundations, principles, and guidelines for the development and application of AI in Brazil.

This was the seventh version of the report published by the Commission. Approved amendments include the following changes:

  • excluding social media algorithms from the list of AI systems considered high-risk;
  • preliminary assessments of AI systems are only mandatory for generative and general-purpose AI systems, with preliminary assessments being optional good practice measures for other AI systems;
  • copyrighted content may only be used in text mining processors for the development of AI systems for research, journalism, archiving, by libraries, museums, and educational organizations;
  • copyrighted material must be obtained legitimately and for non-commercial purposes;
  • citizens are guaranteed the right to explanation and a human review of decisions that have a relevant legal impact; and
  • in the case of AI systems that perform biometric identification, there must be a guarantee of protection against, direct, indirect, illegal, or abusive discrimination.

You can read the press release here and the bill here, both only available in Portuguese

UPDATE (11 December 2024)

Bill passes Senate

The Senate issued, on 10 December 2024, Opinion No. 207 of 2024, approving the bill, following the passage of the bill by the Commission. The Senate rejected the amendments proposed after the bill's passage by the Commission. 

Next steps

The bill will now be sent to the Chamber of Deputies for voting.

You can read the press release here, the opinion here, the latest version of the bill here, and track its progress here, all only available in Portuguese.

UPDATE (23 December 2024)

Senate publishes revised text for AI bill

The Brazilian Senate published, on 19 December 2024, the revised text for the bill, following the Senate's approval of the bill on 10 December 2024.

You can read the revised text here and track the bill's progress here, both only available in Portuguese.