Requires human supervision in AI systems and assigns civil liability for damages from AI usage to the supervisor. Requires state entities to train workers on AI and create specific workers' protection policies.
Analysis summaries, actor details, and coverage mappings were LLM-classified and may contain errors.
This is a binding legislative instrument (Bill of Law) proposed to the Brazilian National Congress with mandatory language and civil liability provisions for non-compliance.
The document has minimal coverage of approximately 3-4 subdomains, with primary focus on human supervision and oversight (5.1, 5.2), governance structures (6.5), and system reliability (7.3). The coverage is concentrated on human-AI interaction risks and basic governance principles rather than technical AI safety or security concerns.
This is a horizontal regulation that applies across all economic sectors in Brazil. The document establishes general principles for AI usage applicable to both private sector entities and government agencies across all industries, with specific provisions for public administration.
The document covers multiple AI lifecycle stages with primary focus on deployment and operation. It addresses design principles, deployment requirements (human supervision), and operational monitoring, but does not explicitly cover data collection, model building, or formal verification/validation processes.
The document uses the general term 'Artificial Intelligence' and 'Artificial Intelligence systems' throughout without defining specific technical categories. It does not distinguish between AI models, systems, or specific types like generative or predictive AI, nor does it mention compute thresholds or open-source models.
The document is explicitly authored and proposed by Senator Styvenson Valentim to the Brazilian National Congress, as indicated in the signature and the legislative format.
The document assigns enforcement responsibilities to government entities at all levels and establishes civil liability mechanisms that would be enforced through the court system.
The document requires government entities to take proactive regulatory action and establishes human supervision requirements, implicitly creating monitoring responsibilities at both governmental and operational levels.
The law applies broadly to all AI usage in Brazil, with specific provisions for government entities and requirements for human supervision across all AI deployments.
8 subdomains (8 Minimal)