On April 26, 2024, Bill No. 1465/2024 establishing principles, guarantees, rights, and obligations for the development, implementation, and application of artificial intelligence in Brazil was introduced to the Chamber of Deputies.

What are the main provisions of the bill?

The bill establishes objectives for artificial intelligence (AI) development and implementation, including:

• respecting human rights and democratic values;
• respecting copyright;
• encouraging economic and sustainable development; and
• protecting freedom of expression.

The bill defines an AI system as a system based on a computational process that operates on different levels of autonomy and uses algorithms for machine learning, natural language processing, or computational visualization to process structured and non-structured data, continuously learning and responding to the environment and making predictions, recommendations, classifications, or decisions to reach specific objectives defined by humans.

Additionally, the bill provides for other definitions, such as 'AI system operator,' 'AI agents,' and 'user.'

Data subject rights

The bill provides for a series of rights of data subjects affected directly or indirectly by AI, including the right to:

• information prior to interacting with the AI system;
• explanation regarding the decision made by the AI system;
• contest decisions or predictions made by AI;
• human involvement in AI systems' decisions;
• non-discrimination;
• privacy and data protection;
• access and portability concerning the data provided to the AI system;
• algorithmic transparency; and
• mechanisms of explicit consent on the use of personal data ensuring that data subjects have clear and easy control over their consent.

Obligations

The bill also provides for obligations for suppliers of AI systems, mainly to:

• establish independent audit mechanisms to evaluate compliance with the applicable legislation and principles;
• put in place a governance structure and internal procedures to ensure the security of AI systems; and
• conduct a risk analysis before making the AI system available on the market, and categorize the risk as excessive, high, or moderate according to the criteria of the bill.

The bill further prohibits the implementation of AI systems considered to represent excessive risk. For high-risk AI systems, the bill provides specific measures that must be taken by suppliers.

Penalties

Finally, the bill provides for civil liability of AI system suppliers and operators for material or non-material and individual or collective damage. Violations of the bill would be punishable with administrative sanctions, such as warnings and fines of up to BRL 100 million (approx. $19.5 million) or, in the case of a company, 2% of the total turnover.

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