Brazil - AI Use Principles (5.051/2019)

Overview

Projeto de Lei nº 5.051/2019 is a short, principles-based bill introduced by Senator Styvenson Valentim to the Federal Senate on 16 September 2019 that seeks to establish the foundational principles for the use of Artificial Intelligence (AI) in Brazil. The bill frames AI as technology "developed to serve people" with the objective of improving human well-being, and sets out a small set of normative principles and public-policy directives rather than technical or sector-specific obligations. The official bill text and its published avulso (single-sheet project) are available in the Federal Senate legislative repository (see Project PDF - Senado Federal) and the consolidated bicameral entry on the Congresso Nacional portal (PL 5051/2019 - Congresso Nacional). The initiative is typical of early-stage AI lawmaking: it sets high-level values—human dignity, protection of privacy, transparency and human supervision—and directs public authorities to promote education, worker retraining and gradual adoption. While not prescriptive, the bill influenced debate and was later processed together with later, more detailed legislative initiatives in the Senate's Temporary Internal Commission on Artificial Intelligence (CTIA), which considered and produced reports consolidating multiple AI bills.

Definitions

The bill itself contains limited definitional content; rather than introducing technical definitions for "AI" it treats the term generically as systems and technologies capable of producing automated or semi-automated outputs. For practical interpretation, the text refers to "systems based on Artificial Intelligence" and to the roles of "supervisor" and "human supervision". Because the bill is principles-based, implementers, regulators and courts would need to rely on supplementary definitions adopted in implementing instruments, regulatory guidance, sectoral rules, or broader statutory definitions (for example, definitional work produced by the CTIA and referenced reports on AI governance). In parallel, national frameworks such as the Brazilian Data Protection Law (Lei Geral de Proteção de Dados – LGPD, Law No. 13.709/2018) supply operational definitions relevant to personal data processing by AI systems.

Governance and Institutional Framework

PL 5051/2019 does not create a new centralized AI regulator; its governance approach is distributed and principle-driven. The bill tasks the Union, states, Federal District and municipalities with carrying out directives to promote education, worker protection and gradual adoption. Much of the formal deliberation on the bill occurred in the Federal Senate, notably in the Commission on Science, Technology, Innovation, Communication and Informatics (CCT) and later in the Senate's Temporary Internal Commission on Artificial Intelligence in Brazil (CTIA). The CTIA produced extensive reports and a consolidating substitute text that examined multiple AI bills together; these proceedings and reports are available on the Senate website (for example, CTIA work and reports are published via the Senate activity pages and news releases such as the CTIA report and related PDFs available on the Senate portal). The bill therefore relies on legislative committees and inter-agency coordination, rather than a single executive regulator, to translate principles into practice. Relevant existing agencies likely to be engaged in implementation or enforcement include sectoral supervisors and the national data protection authority (Autoridade Nacional de Proteção de Dados – ANPD), although the bill does not delegate explicit regulatory powers to them.

Key Focus Areas

PL 5051/2019 emphasizes five core areas: (1) human-centric values and fundamental rights – respect for human dignity, freedom, democracy and equality; (2) human rights, pluralism and diversity – ensuring inclusive development and avoiding discriminatory impacts; (3) privacy and data protection – explicit guarantee of privacy and personal-data safeguards consistent with Brazil's existing LGPD; (4) transparency, reliability and auditability – systems should be transparent and auditable so outcomes can be understood and reviewed; and (5) human supervision – AI decision-support systems must remain auxiliary to human decision-making and require supervision appropriate to the type, gravity and implications of decisions. The bill also sets public-policy directives: promote education for cognitive, emotional and economic development compatible with AI; enact policies to protect and reskill workers; encourage gradual adoption of AI technologies; and adopt proactive regulation of AI applications. In short, the bill champions a rights-respecting, human-centred and precautionary approach that balances innovation with safeguards.

Implementation Framework

The text provides a limited implementation framework: it is declaratory and aspirational rather than prescriptive. Article 5 lists directives for the public sector (education, worker qualification, gradual adoption and proactive regulation). Article 4 establishes that systems based on AI are always auxiliary to human decision-making, while the statutory clauses on supervision require that the form of human oversight be compatible with the seriousness of the decisions being supported. Article 2 enumerates the normative principles and Article 6 instructs public entities using AI to seek quality and efficiency in public services. Implementation would therefore require secondary measures: normative regulations, sectoral rules, supervisory guidance (potentially from ANPD in relation to data protection), procurement rules for public sector AI use, technical standards for transparency and auditability, and capacity-building programs in government. The CTIA and the reports it produced can serve as sources for interpretive guidance and potential models for delegated instruments (see CTIA documentation and the Senate reports cited on the official Senate pages).

Monitoring and Evaluation

PL 5051/2019 does not establish a formal monitoring body or metrics, but it anticipates active regulatory attention by urging proactive regulation and quality/efficiency goals for public sector AI. Monitoring and evaluation would logically be handled through: (a) legislative committee reviews and reporting; (b) sectoral supervisors and audit institutions assessing public-sector AI deployments; (c) data protection oversight by the ANPD for privacy-related impacts; and (d) periodic reviews under any implementing regulations. The Senate CTIA's hearings, published minutes and consultative processes illustrate an open model of public consultation and iterative scrutiny that could be adapted for ongoing monitoring (see the CTIA news and reports on the Senate site).

Penalties, Liability, and Appeals

The bill assigns civil liability for damages resulting from the use of AI systems to the human "supervisor" of the system, making that person responsible for harms attributed to system operation. The text does not create administrative sanctions, criminal penalties, or a specialized enforcement regime; it thus relies on existing civil liability frameworks and sectoral enforcement mechanisms. Because the bill lacks specific procedural or sanctioning provisions, appeals, remedies and procedural safeguards would depend on general administrative and civil law procedures, including judicial review, administrative complaint mechanisms available in each sector, and LGPD enforcement where personal-data violations occur. The absence of express penalties beyond civil liability was a factor in subsequent legislative consolidation and the preparation of more detailed substitute texts by CTIA and other actors.

Relationship to Other Instruments

PL 5051/2019 must be read alongside Brazil's Lei Geral de Proteção de Dados (LGPD, Law No. 13.709/2018) which governs personal-data processing and is explicitly referenced by commentators as complementary to the bill's privacy protections. In the Senate, PL 5051/2019 was later processed jointly with other AI bills (e.g., PL 5691/2019, PL 21/2020, PL 872/2021, PL 2338/2023 and PL 3592/2023) within the CTIA, and aspects of its principles were taken into account, amended or declared prejudiced during consolidation processes documented in Senate reports. The Congresso Nacional bicameral portal records the project's bicameral processing and related materials.

International Alignment

The bill's principles track international AI governance trends emphasizing human-centric AI, human oversight, privacy protection and transparency as set out in documents from multilateral and regional bodies. While PL 5051/2019 is not a technical mirror of any single foreign regulatory instrument, its emphasis on rights, privacy and human supervision aligns with approaches seen in the EU's AI policy discourse, OECD AI Principles, and other national AI strategies. The CTIA's hearings included analysis of international models and invited experts to discuss regulatory alternatives, furthering alignment and comparative analysis (see CTIA meeting records on the Senate site).

Implementation Timeline

Event	Date
Introduction and first reading (Senate)	2019-09-16
Processing in Senate commissions and publication of project PDF	2019 (published avulso)
CTIA consolidation and reporting activity	2023-2024 (reports and hearings)
Relator reports and complementations (CTIA)	2024-06-07 to 2024-12-05
Declared prejudiced / archived	2024-12-10

Compliance Checklist

Requirement	Notes
Human-centred design	Adopt AI systems with priority on human dignity and rights (Article 2).
Human supervision	Ensure appropriate human oversight compatible with the decision's gravity (Article 4 §1).
Liability assignment	Identify a human supervisor who will hold civil liability for damages (Article 4 §2).
Privacy and data protection	Ensure compliance with privacy protections and relevant data protection law (Article 2 III).
Transparency and auditability	Design systems to be auditable and dependable (Article 2 IV).
Public-sector quality	Public AI applications must seek quality and efficiency (Article 6).
Education and workforce policies	Implement worker protection and qualification measures (Article 5 I–II).

Sources and References

Source	Type
PROJETO DE LEI N° 5051, de 2019 (avulso / full text - Senado Federal)	Primary Source
PL 5051/2019 - Senate activity page (Senado Federal)	Primary Source
PL 5051/2019 - Congresso Nacional (bicameral record)	Primary Source
PL 5691/2019 and related AI bills - Senate consolidated records (CTIA)	Primary Source