Brazil AI Plan 2024

Overview

The Plano Brasileiro de Inteligência Artificial (PBIA) 2024-2028 is Brazil’s national strategic plan for developing, making available and using artificial intelligence to address major social, economic, environmental and cultural challenges. Coordinated by the Ministry of Science, Technology and Innovation (MCTI) under the strategic direction of the National Council for Science and Technology (CCT), the plan presents AI as an instrument for social welfare, technological sovereignty, sustainable development and better public services. It expressly seeks to protect individual and collective rights, promote inclusion, defend democracy and information integrity, protect workers, and strengthen national autonomy.

The plan is a policy and investment instrument rather than an AI statute. It does not independently impose legally enforceable system-provider obligations, licensing requirements, administrative fines, or civil-liability rules. Its operational significance lies in directing federal programmes, public funding, institutional coordination and measurable implementation goals. The PBIA combines immediate-impact initiatives—intended to demonstrate tangible results, often within 12 months—with structural initiatives intended to create lasting national capacity in computing, data, research, workforce development, public administration and business innovation.

Definitions

For PBIA purposes, artificial intelligence comprises models, algorithms, techniques and methodologies implemented as computational systems that generate outputs such as predictions, classifications, recommendations and decisions through learning processes, commonly using substantial volumes of data. Such systems may operate with differing degrees of autonomy, ingest human- or machine-generated inputs, form abstract models through training, and infer, generate information, act, or support decision-making. The definition includes specialised and general-purpose systems, as well as generative AI that creates or significantly modifies text, images, audio, video and software code.

The PBIA’s glossary also frames operational concepts relevant to responsible AI. “Responsible AI” is the collection of methods and processes used to develop, use and deploy ethical, reliable and socially beneficial systems. “Traceability” means the ability to track and document data and AI decisions throughout the lifecycle. “Transparency” concerns the extent to which persons can understand the functioning of models, algorithms and systems. “Reliable AI” encompasses systems that are fair, safe, secure, resilient, accountable, transparent and interpretable, while “sustainable AI” refers to environmentally responsible and socially just AI use.

Governance and Institutional Framework

The PBIA was developed under the CCT’s strategic guidance and coordinated by the MCTI, with support from the Centro de Gestão e Estudos Estratégicos (CGEE). Its development involved a working group established by MCTI Ordinance No. 8,251 of 7 June 2024, government bodies, research institutions, industry, civil society and other stakeholders. The plan describes a participatory model in which the State, academia, private sector and civil society should collaborate in its design, execution, monitoring and assessment.

For plan-wide governance, the PBIA assigns coordination to the Comitê Interministerial para a Transformação Digital (CIT Digital), a consultative collegiate body established by Decree No. 12,308/2024 to advise the President on digital-transformation public policies. In the federal public-service axis, the Núcleo de Inteligência Artificial do Governo Federal is to be coordinated by the Secretariat of Digital Government (SGD), with initial participation by MGI, MCTI, Enap, Serpro and Dataprev. Its stated functions include knowledge-sharing, support for common standards and experimentation, workforce development, and monitoring AI adoption in government.

Key Focus Areas

The PBIA organizes structural measures across five axes: infrastructure and AI development; dissemination, education and capacity building; AI for improved public services; AI for business innovation; and support for AI regulation and governance. The investment envelope for 2024-2028 is R$23.03 billion. Of this amount, R$435.04 million is allocated to immediate-impact actions; R$5.79 billion to infrastructure and development; R$1.15 billion to education and skills; R$1.76 billion to public services; R$13.79 billion to business innovation; and R$103.25 million to regulatory and governance support.

Priority infrastructure measures include upgrading the Santos Dumont supercomputer so it ranks among the world’s five largest systems within five years, expanding regional high-performance computing centres, connecting national supercomputing facilities through high-speed networks, supporting AI infrastructure in scientific and technological institutions, and developing a Brazilian software stack and Portuguese-language LLMs based on curated national datasets. The plan also emphasizes energy efficiency, renewable energy, national data resources, cybersecurity, reduced technological dependence and sustainable data-centre development.

Implementation Framework

The PBIA uses a dual implementation model. Immediate-impact actions target specific, well-delimited public-interest problems using available or advanced AI technologies and existing data sources. These initiatives are expected to have clear indicators, measurable benefits, potential for scaling and replication, and direct relevance to affected populations. Listed sectors include health, agriculture and livestock, environment and climate, industry and services, education, social development, public-service management, labour and employment, and defence, public security and cybersecurity.

Structural actions are intended to build medium- and long-term capacity. Each action is generally described through its purpose, challenge, measurable target, expected impact, 2024-2028 budget and funding source, and components. Examples include a sovereign government cloud; a federal AI platform; national data cataloguing, governance and reuse measures; AI cybersecurity tools for government; workforce and teacher-training programmes; data centres powered by renewable energy; funding for AI startups and industrial projects; and programmes for Brazilian responsible-AI guidance, algorithmic transparency and AI-governance networks.

Monitoring and Evaluation

Monitoring is a core design element. Immediate-impact actions are expected to use output, outcome and social-impact metrics, such as implementation coverage, user feedback, service quality, cost reduction, mortality or morbidity indicators where appropriate, educational progression, employment outcomes, fraud detection and environmental results. The plan requires that actions address data quality, adequacy, privacy, security and potential bias. It also states that actions should incorporate feedback from target populations whenever possible and consider financial sustainability beyond initial implementation.

Structural measures are to be periodically reviewed and adjusted in light of achieved progress, technological developments, global conditions and lessons learned. The federal AI nucleus is tasked with monitoring government AI use and conducting annual surveys of adoption by federal bodies. The plan also provides for the consolidation of the Observatório Brasileiro de Inteligência Artificial (Obia) as a national intelligence platform on AI, including continuous development of indicators. This monitoring approach is policy-oriented and adaptive; it does not establish a statutory inspection or sanction regime for private AI operators.

Penalties, Liability, and Appeals

The PBIA does not establish AI-specific penalties, administrative sanctions, appeals procedures, private rights of action or a dedicated liability regime. It recognizes that responsibility for AI-generated outcomes may be technically difficult to allocate because multiple participants in the AI value chain can contribute to a result. The plan identifies accountability, auditability, transparency, traceability, privacy, cybersecurity, consumer protection, intellectual property and copyright as important policy concerns, especially in high-risk domains such as health, public security and autonomous mobility.

Rather than creating direct enforcement duties, the plan supports development of governance mechanisms and a suitable legal framework. Its fifth axis includes a Programme to Support Improvement of the AI Regulatory Framework, with work on impact studies, risk-assessment methodologies and a National Centre for Algorithmic Transparency and Trustworthy AI. Existing Brazilian law remains relevant, including the General Data Protection Law (LGPD) for personal-data processing, the Access to Information Law, open-data rules and federal public-sector data-sharing rules. Any binding duties or remedies must therefore be derived from applicable legislation or later regulations, not from the PBIA alone.

Relationship to Other Instruments

The PBIA expressly aligns AI development with Brazil’s broader governmental policies. It highlights the Nova Indústria Brasil (NIB), under which AI is intended to help advance industrial productivity and the programme’s missions, including sustainable agro-industrial chains, health, digital industrial transformation, bioeconomy, decarbonisation and national sovereignty. It also aligns sustainable AI infrastructure with the Plano de Transição Ecológica. The plan treats the country’s predominantly clean electricity matrix as a strategic advantage, while recognizing that computing infrastructure may create significant energy, water and electronic-waste pressures.

For governance, the PBIA situates AI within existing rules on information access, open data, personal-data protection and federal data sharing. It specifically identifies Law No. 12,527/2011, the LGPD (Law No. 13,709/2018), Decree No. 8,777/2016 on the Federal Executive’s Open Data Policy, and Decree No. 10,046/2019 on federal data-sharing governance. The plan also notes ongoing debate on Bill No. 2,338/2023 as part of Brazil’s evolving AI regulatory landscape, but it does not claim that the bill has been enacted or itself substitute for legislation.

International Alignment

International cooperation is one of the PBIA’s ten foundational premises. The plan calls for partnerships that are fair, mutually beneficial and compatible with Brazilian technological and data sovereignty. It seeks to expand Brazilian capabilities while supporting more equitable global access to AI knowledge, infrastructure and opportunity. The plan gives particular attention to collaboration with countries in Latin America, the Caribbean and Africa, and includes a programme supporting collaborative research and development projects with institutions in those regions.

The PBIA also presents Brazil as a prospective voice for developing countries in international AI governance. It proposes a network of researchers and technical experts to strengthen Brazilian participation in international AI debates, initiatives, plans, normative processes and resolutions. The plan’s stated international orientation includes protection of democracy and information integrity, cooperation on standards and governance, and sharing of Brazilian research infrastructure with developing countries. These ambitions are strategic objectives, not international legal commitments or mutual-recognition arrangements.

Implementation Timeline

Compliance Checklist

Sources and References