IBGIA - Instituto Brasileiro de Governança em Inteligência Artificial
Complete Guide

AI Governance in Brazil

Understand what Artificial Intelligence governance is, why Brazil needs it, what the regulatory framework says, and which principles guide the responsible development of AI.

What is AI Governance?

Artificial Intelligence Governance is the set of policies, regulations, standards, practices and institutional mechanisms that guide the development, deployment and responsible use of AI systems. It goes far beyond simple legal compliance: it encompasses technical, ethical, organizational and social dimensions. If you encounter unfamiliar terms in this guide, consult our AI governance glossary.

In practice, AI governance means ensuring that algorithmic systems make decisions transparently, fairly and auditably. It involves creating accountability structures within organizations, establishing risk assessment processes, implementing human oversight mechanisms and ensuring that people's fundamental rights are protected.

Unlike terms such as AI regulation (which refers specifically to legal norms), governance is a broader concept that also includes self-regulation, sectoral codes of conduct, maturity frameworks, technical standards and social participation mechanisms. Good AI governance integrates all these elements into a coherent ecosystem.

For organizations, implementing AI governance is not just a matter of anticipatory compliance with legislation such as PL 2338/2023 — it is a competitive advantage. Companies with mature governance practices reduce reputational risks, avoid algorithmic discrimination, build trust with their clients and position themselves better with investors and partners.

Why Does Brazil Need AI Governance?

Largest AI market in Latin America

Brazil is the country with the highest AI adoption in Latin America. According to recent research, over 40% of large Brazilian companies already use some form of AI in their processes. Banks, retailers, healthcare operators and even government agencies incorporate algorithms into decisions affecting millions of people. This level of adoption requires a proportional governance framework.

AI already present in public services

Brazil already uses AI in several critical areas: the Judiciary employs systems like Victor (STF) and Elis (TJMG) for case triage; the financial system uses algorithmic scoring for credit decisions; and there is growing use of facial recognition in public security. These applications impact fundamental rights and demand adequate oversight.

Gap between adoption and regulation

While AI adoption advances rapidly, Brazil still lacks a specific legal framework for AI. The LGPD covers personal data aspects but does not address specific algorithmic governance issues. PL 2338/2023 seeks to fill this gap, but its approval is still under consideration. Responsible organizations do not wait for the law to implement best practices.

The Brazilian scenario presents unique challenges: social inequality amplifies the risks of algorithmic discrimination; cultural and regional diversity requires AI models to be sensitized to local contexts; and heterogeneous digital infrastructure creates access barriers that can be worsened by automated systems. AI governance in Brazil cannot simply copy European or North American models — it needs to be built from our own realities.

Brazilian Regulatory Framework

Brazil's AI regulatory ecosystem consists of existing legislation, bills under consideration and the actions of regulatory bodies.

PL 2338/2023 — AI Legal Framework

Under consideration in the Chamber of Deputies

The main bill for comprehensive AI regulation in Brazil, authored by Senator Rodrigo Pacheco. Developed by the Senate's Commission of Jurists (CJIA), it adopts a risk-based approach inspired by the EU AI Act, with adaptations to the Brazilian context.

The bill establishes: classification of systems by risk level (unacceptable, high, limited, minimal); rights of affected persons (explanation, contestation, human review); transparency obligations and algorithmic impact assessment; and creation of a specific regulatory body.

LGPD and the Intersection with AI

Law 13,709/2018 — In force

The General Data Protection Law is already the main legal instrument applicable to AI systems in Brazil. Article 20 guarantees the right to review automated decisions; the law requires a legal basis for processing data used in model training; and the principles of purpose, adequacy and necessity apply directly to AI development.

However, the LGPD was not designed specifically for AI. There are gaps in areas such as algorithmic transparency, social impact assessment (beyond personal data impact), governance of generative AI models and liability for damages caused by autonomous systems. The AI Legal Framework aims to complement and harmonize these rules.

The Role of ANPD

National Data Protection Authority

ANPD plays a central role in Brazil's AI regulatory ecosystem, given the deep relationship between AI and personal data. Since 2023, the authority has published guides and technical notes on AI, including analyses on generative AI and data processing for model training.

There is an active debate about whether ANPD should also assume the role of AI regulatory body envisioned in PL 2338/2023, or whether a new entity should be created. The issue involves considerations about institutional capacity, independence, technical expertise and potential conflicts of competence with other sectoral bodies (BACEN, ANS, CADE).

International References

AI governance is a global topic. Understanding the approaches of other countries and international organizations is essential to position Brazil in this debate.

EU AI Act (European Union)

Risk-based approach

The world's first comprehensive AI legislation. Adopts a risk-based approach with four levels (unacceptable, high, limited and minimal), prohibits practices such as social scoring and mass biometric surveillance, and creates specific obligations for generative AI. Entered into force in August 2024 with progressive implementation until 2027.

Recommendation on the Ethics of AI (UNESCO)

193 signatory countries

Adopted by 193 countries in November 2021, it is the first global normative instrument on AI ethics. Establishes values (human dignity, well-being, diversity), principles (proportionality, safety, fairness, transparency) and policy action areas. Brazil is a signatory.

OECD AI Principles

40+ adopting countries

Adopted in 2019 and updated in 2024, they serve as a reference for public policy in more than 40 countries. They include five principles: inclusive growth, human-centered values, transparency, robustness/safety and accountability. The OECD AI Policy Observatory monitors global implementation.

Executive Orders on AI (USA)

Focus on frontier AI

The Executive Order on AI Safety (October 2023) established safety requirements for frontier models, NIST standards for risk assessment and transparency obligations for developers. Although revoked in January 2025, it influenced the global debate on frontier AI governance.

Principles of Responsible Governance

Regardless of the regulatory framework, there is a set of principles widely recognized by the international community as fundamental to the responsible development of AI. These principles are the foundation of any governance program.

Transparency

AI systems must be understandable. Organizations need to explain how their algorithms work, what data they use and how they reach decisions that affect people. Transparency is a prerequisite for public trust and effective oversight.

Accountability

There must always be a person or organization responsible for the decisions and impacts of an AI system. Accountability requires traceability of decisions, documentation of processes and clear mechanisms for damage repair.

Fairness and Non-Discrimination

AI systems must not perpetuate or amplify existing prejudices and inequalities in society. This requires rigorous bias testing, diversity in training data and continuous monitoring of results across different population groups.

Privacy and Data Protection

The development and use of AI must respect privacy rights and personal data protection standards, especially the LGPD. Techniques such as privacy by design, data minimization and anonymization are essential.

Safety and Robustness

AI systems must be safe, reliable and resilient to attacks and failures. This includes rigorous testing before deployment, continuous monitoring in production, contingency plans and protection against adversarial manipulation.

Human Oversight

Humans must maintain meaningful control over AI systems, especially in high-impact decisions. The 'human-in-the-loop' principle ensures that critical decisions are not entirely delegated to algorithms without human review.

The Role of IBGIA

Building AI governance from the Brazilian reality

The IBGIA — Brazilian Institute for AI Governance — is a nonprofit civil association founded with the mission of promoting responsible AI governance in Brazil and Latin America. We act as a bridge between different sectors of society: government, business, academia and civil society.

Our work is organized into four areas. In the area of assessment and compliance, we develop maturity frameworks and diagnostic tools. In the area of education and advocacy, we produce technical content and actively participate in the regulatory debate. In the area of consulting, we support organizations in implementing governance practices. And in the area of research, we contribute to advancing knowledge about AI governance in the Brazilian and Latin American context — see our publications.

We believe that AI governance in Brazil cannot be imported from other contexts without adaptation. It needs to be built from our own social, economic and cultural realities — with technical rigor, institutional independence and commitment to the public interest.

Deepen your knowledge

Governance by Sector

Each sector faces unique challenges in AI adoption. Explore our in-depth guides for sectoral governance.

Receive AI governance updates

Subscribe to the IBGIA newsletter and receive analyses, frameworks and updates on artificial intelligence governance in Brazil.

Exclusive AI governance insights

Get weekly analyses, regulatory alerts and practical frameworks on AI governance in Brazil — free.

Want to contribute to AI governance in Brazil?

Join IBGIA and be part of building responsible, ethical artificial intelligence aligned with the public interest.