Brazil's AI Act: A Comprehensive Framework for AI Governance
Brazilian National Congress
Plenário Ulysses Guimarães
Image: Brazilian National Congress Chamber
Introduction
Brazil is on the cusp of enacting one of the world's first comprehensive artificial intelligence laws outside of Europe. In December 2024, the Brazilian Federal Senate approved Bill No. 2338/2023, commonly referred to as the Brazil AI Act, which sets out a legal framework for the development and use of AI in Brazil. This landmark bill – years in the making – is inspired by global models like the EU's AI Act but tailored to Brazil's context and values. It adopts a risk-based, human rights-centric approach to AI governance, aiming to protect fundamental rights while fostering innovation.
As the bill moves through Brazil's House of Representatives in 2025, it has drawn intense debate from policymakers, industry, and civil society, underscoring its significance as a potential model for Latin America and other emerging economies.
Background: From Principles to a Risk-Based Law
Brazil's journey toward AI regulation began with high-level principles and evolved into the detailed risk-based framework now on the table. Early efforts date back to 2020–2021, when Brazil's Congress considered a more general AI principles bill (PL 21/20) that passed the Lower House but faced criticism for being too narrow and vague. Learning from that experience, the Senate took charge: in 2022, Senate President Rodrigo Pacheco convened a committee of jurists to study AI governance. Their December 2022 report recommended a risk-tiered regulatory model, aligning with the emerging international consensus that AI rules should focus on higher-risk uses.
Building on these recommendations, Pacheco introduced Bill 2338/2023 in May 2023. Over the next 19 months, a special Senate AI committee held dozens of meetings and public hearings to refine the bill. Intense negotiations took place, balancing the ambition to lead in AI regulation with concerns from industry and other stakeholders. By the time the Senate voted in December 2024, the draft had been "watered down" in some areas under lobbying pressure, yet it still "remains protective of fundamental rights" according to Brazil's data protection watchdogs.
Key Legislative Debates
- •Information Integrity: The concept requiring AI systems to promote reliable, truthful information was championed by the government but met with resistance from tech lobbyists who feared it could justify censorship.
- •AI Regulator: An initial idea to create a new AI oversight agency was dropped mid-2024 in favor of empowering existing bodies to coordinate AI governance.
Key Provisions of the Brazil AI Act
Risk Categories
The law defines tiers of AI system risk with corresponding rules:
Excessive-Risk AI (Prohibited)
AI systems that manipulate human behavior to cause harm, exploit vulnerable groups, engage in illegitimate social scoring, or enable real-time remote biometric identification in public for surveillance purposes. Such uses are banned due to their threat to health, safety, fundamental rights, or democratic processes.
High-Risk AI (Regulated)
Applications with significant implications for society or individuals, ranging from AI in critical infrastructure to algorithms used in education, employment, essential services access, healthcare, biometric identification, and law enforcement. These systems face strict oversight and compliance requirements.
Lower-Risk AI
AI systems not falling into the above tiers are subject to general obligations but fewer onerous rules.
Obligations on AI Actors
Brazil's draft law casts a wide net in assigning responsibilities, covering the full AI value chain:
Developers
Any party who builds an AI system and places it on the market or uses it under their brand
Distributors
Those who make an AI system available to others (e.g., app stores, cloud platforms)
Operators
End-user entities who deploy or use the AI system in their processes
Unlike the EU's approach, which concentrates many obligations on providers of high-risk AI, Brazil's bill spreads compliance duties across all these actors. Over half of the 56 obligations in the Senate-approved text apply equally to providers, distributors, and operators.
Pre-Market Risk Assessment
Every AI system must undergo preliminary risk classification before launch. High-risk systems trigger mandatory Algorithmic Impact Assessments (AIAs) that evaluate potential harm, bias, and mitigation measures.
Transparency and User Rights
Developers must disclose AI interactions and provide explanations for high-risk systems. Affected individuals have channels to appeal automated decisions and obtain human review.
Intellectual Property and Training Data
The law includes robust copyright protections, requiring developers to disclose copyrighted material use and compensate rights holders – provisions more far-reaching than the EU's AI Act.
Comparison with the EU AI Act and Global Standards
Key Differences from EU Approach
Regional Impact: A New Benchmark for Latin America
Brazil is the largest economy in Latin America, and its bold step on AI regulation is poised to reverberate across the region. As Eduardo Levy Yeyati observed, Latin American countries face a choice: either develop AI governance on their own terms or become "a regulatory colony" of external powers like Silicon Valley or Brussels. In this context, Brazil's AI Act serves as a homegrown benchmark that could inspire neighboring countries to craft their own AI laws aligned with regional values.
Countries Following Suit
- • Chile: Drafted AI legislation based on transparency and fairness
- • Colombia, Peru, Paraguay: Working on AI governance proposals
- • Argentina: Growing momentum through congressional hearings
Regional Cooperation
ECLAC's 2024 Digital Agenda calls for regional coordination on AI standards. Brazil's framework might form the backbone of shared standards across Latin America.
Implications for Tech Companies
For U.S. Tech Giants
Silicon Valley giants face new compliance challenges, including:
- • Brazil-specific Algorithmic Impact Assessments
- • Transparency features for AI-curated content
- • Internal governance teams to monitor bias
- • Potential retraining of models to respect opt-out rights
For Latin American Startups
The Act presents both opportunities and challenges:
Opportunities
- • Level playing field with ethical AI norms
- • Market differentiation through compliance
- • Growth of AI compliance industry
Challenges
- • Compliance costs for resource-constrained firms
- • Potential barriers to entry
- • Need for specialized expertise
Outlook: Brazil's AI Act on the Global Stage
Brazil's AI Act represents a bold attempt by a middle-income democracy to shape the trajectory of AI within its borders and beyond. As of mid-2025, the bill awaits final approval in Brazil's Chamber of Deputies and the President's signature, steps which the Lula administration prioritizes within this legislative year.
For policy experts and industry leaders, Brazil's experiment is incredibly valuable. It offers a preview of how AI governance might play out in democracies outside the West, where resources are more limited but the commitment to rights is strong. If Brazil succeeds in passing and implementing this law effectively, it could position itself as a global reference point for AI regulation alongside the EU.
Key Takeaways
The Brazil AI Act is a groundbreaking policy initiative setting a high-water mark for AI legislation in the Global South. Its downstream effects are already being felt:
- •Lawmakers across the Americas are paying attention
- •Corporations are adapting their strategies
- •Latin America is taking an influential seat at the global AI governance table
Sources
• White & Case – AI Watch: Global Regulatory Tracker – Brazil (June 2025)
• Tech Policy Press – "Brazil's AI Law Faces Uncertain Future as Big Tech Warms to Trump" (Feb 2025)
• Data Privacy Brasil – "The Artificial Intelligence Legislation in Brazil: Technical Analysis…" (Dec 2024)
• Covington & Burling – "Brazil's Digital Policy in 2025: AI, Cloud, Cyber…" (Feb 2025)
• Americas Quarterly – "Regulating AI on Latin America's Terms" (June 2025)
• Courthouse News – "Brazil's justices clear road for tech platform liability…" (June 2025)
• Rest of World – "Brazil's push for comprehensive AI regulation" (Jan 2025)
• Pulitzer Center (Núcleo) – "How Conflicts… Shaped Brazil's AI Regulation Proposal" (Dec 2024)
• OpenDemocracy – "How Trump's trade war with Brazil serves Big Tech's interests" (July 2025)