Brazil: Introduced Artificial Intelligence Act (PL 1465/2024) including design and testing standards

On 26 April 2024, a Bill on Artificial Intelligence (PL 1465/2024) was introduced to the Brazilian Chamber of Deputies. The Act defines an AI system as one that relies on computational processes functioning at various levels of autonomy, using algorithms in machine learning, natural language processing, or computer vision. These systems handle both structured and unstructured data, continually learning from and reacting to their surroundings. They are capable of making predictions, recommendations, classifications, or decisions to achieve specific goals set by humans. The Act establishes objectives for AI development and implementation, including but not limited to respecting intellectual property rights, human rights, democratic values, and freedom of expression. Article 7 requires that providers of AI systems establish mechanisms for independent audits. These audits should evaluate the systems' compliance with ethical principles, existing legislation, and guidelines set by the competent authority. Article 8 mandates AI agents to establish comprehensive governance structures and internal processes to ensure the security of AI systems and protect the rights of individuals affected by these technologies. Governance measures must include transparency in the use of AI systems, the adequacy of data management to prevent potential discriminatory biases, the adherence to data protection laws, emphasising privacy by design and default, and the robustness of information security measures from the design phase through to the operational phase of the systems.