Brazil and Peru offer contrasting models for AI governance in Latin America

Brazil and Peru are taking different paths toward AI governance, offering lessons for Latin American countries building regulatory institutions. An Atlantic Council analysis says laws must be matched by clear authority, operational capacity and accountability.
- Brazil is strengthening the ANPD while its general AI framework remains under consideration.
- Peru assigned AI regulatory authority to the SGDT through legislation passed in 2023.
- Brazil’s Superior Electoral Court issued fourteen AI-related resolutions ahead of the 2026 elections.
- Peru divides enforcement responsibilities among data protection, competition and police authorities.
- The Atlantic Council says AI laws must be matched by institutional capacity and accountability.
Two different routes to AI regulation
Brazil is strengthening the institutions that could oversee AI while its general legal framework remains under consideration by a special committee in the Chamber of Deputies. The country’s Senate approved a bill on December 10, 2024, covering the use, development and deployment of AI, but no vote date has been confirmed for the broader framework.
A proposal attached to the bill would expand the role of Brazil’s National Data Protection Authority, known as ANPD, making it the default AI regulator alongside sectoral regulators. Separately, a law enacted on February 25, 2026, made the ANPD a full regulatory agency with greater technical, administrative and financial authority.
Peru has moved further in formalizing its framework. Legislation passed in 2023 established principles covering data privacy, national AI talent and cooperation on AI development. It assigned regulatory authority to the existing Secretariat of Government and Digital Transformation, or SGDT. A supreme decree issued in 2025 approved implementing regulations, including financing provisions, accountability mechanisms, implementation periods and divisions of responsibility among public institutions.
Brazil builds capacity before a general AI law
The Atlantic Council presents Brazil as an example of institutional development before comprehensive AI legislation is settled. The Organization of American States’ Inter-American Framework on Data Governance and Artificial Intelligence, or MIGDIA, offers another tool. Developed by multisector stakeholders and governments, it provides guidance that countries can adapt while their AI laws are still being drafted.
Brazil’s Superior Electoral Court has also acted as an interim regulator in a specific area. Ahead of the 2026 Brazilian elections, it issued fourteen resolutions governing the electoral process, including restrictions on AI use. The resolutions barred AI systems from recommending candidates or signaling voting preferences, and required disclosure of AI-generated or digitally manipulated campaign content. The court could enforce the rules through suspension of profiles.
The analysis says existing judicial institutions could take on similar, limited regulatory roles in countries that do not yet have designated AI regulators, particularly for risks within their established areas of expertise.
Brazil’s experience also highlights the need for locally tailored regulation. Its first AI proposal was criticized for insufficient inclusion of Indigenous voices and for failing to address structural racism in public security. São Paulo’s Smart Sampa surveillance pilot project faced scrutiny from civil society after concerns that it disproportionately profiled black individuals. These cases underscore the role of civil society participation, oversight and accountability in shaping regulation.
Peru focuses on implementation and enforcement
Peru’s challenge is less about assigning formal authority and more about ensuring that institutions can enforce the framework. The SGDT leads implementation, while other public bodies handle enforcement under its regulatory umbrella. The rules allow the SGDT to consult competent authorities when it encounters a regulatory breach.
Responsibilities are divided according to the type of violation and sector involved. The National Data Protection Agency handles AI-related personal data breaches. The National Institute for the Defense of Competition and Protection of Intellectual Property addresses industry competitiveness and consumer protection. The National Police specializes in cybercrimes.
Dilmar Villena, director of Hiperderecho, warned the authors that institutions could have limited capacity to oversee emerging AI concerns without additional resources. “They risk turning into hollow governing bodies that promote norms without enforcement mechanisms,” Villena said, according to the Atlantic Council analysis.
The report also points to Peru’s digital government indicators, which track digital government initiatives and institutional capabilities. They do not yet measure whether AI implementation itself is working, but they can show where administrative capacity is weak and provide a basis for future readiness assessments.
Why institutional readiness matters
The Atlantic Council’s comparison suggests that Latin American countries do not need to follow a single model. Brazil is developing regulatory authority before its general framework is finalized, while Peru has assigned authority through legislation and is beginning to apply its rules.
The shared lesson is that regulation alone does not ensure effective governance. Governments also need clear divisions of responsibility, sufficient personnel, expertise and funding, along with mechanisms for transparency, review, participation and correction when systems fail. Countries that address gaps in authority, operational capacity and accountability, while incorporating government, civil society and technologists, will be better positioned to implement responsible AI practices.
Atlantic Council (AI): Lessons from Brazil and Peru on how to build AI-ready institutions across Latin America


