In the Markets: Brazil
7 min read
2024-09-23
Paulo Brancher
Partner, Mattos Filho

executive summary

Gallery

No items found.
Need this in PowerPoint?
Enter your email to request the file.
Thanks! We'll send the file to your email shortly.
Something went wrong while submitting the form. Get in touch with 20Minds via info@twentyminds.com

article

Sample

AI is one of the most significant technologies of our time. In recent years, the debate on AI-related regulation and the risks posed by AI applications has matured significantly in Brazil, reflecting recent changes in the AI legal landscape.

How has the AI investment landscape evolved in Brazil?

While not yet reaching the scale of the U.S. or China, Brazil’s AI investment landscape is growing, with many startups seeking venture and growth capital.

Key areas of investment include:

  • Natural Language Processing (NLP): Companies like Semantix Inc. are developing Portuguese-based NLP solutions for customer service, chatbots, and sentiment analysis.
  • Computer Vision: Startups like Tractian utilise AI for predictive maintenance and industrial automation, while others focus on applications in agriculture and healthcare.
  • Fintech: AI is being employed in fraud detection, credit scoring, and personalised financial recommendations, with companies like Nubank and Credit as serving as leading examples.

While not yet reaching the scale of the U.S. and China’s respective LLM markets, Brazil is also witnessing growing interest in this technology. For example, the Brazilian government has reportedly engaged OpenAI to reduce litigation costs by using AI to expedite the screening and analysis of lawsuits. The system aims to flag necessary actions before final decisions, identify trends, and suggest strategic areas for the Solicitor General’s Office (AGU). Microsoft will reportedly deliver these AI services via its Azure cloud-computing platform1.

Recent AI Investments in Brazil

Are there policies to help develop the Brazilian AI sector?

Brazil’s Ministry of Science, Technology, and Innovation (Ministério da Ciência, Tecnologia e Inovação, MCTI) has launched programs like the Brazilian Strategy for Artificial Intelligence (EBIA) to support AI research and development. The strategy foresees various measures, including making government data open-source, developing educational and awareness campaigns, and encouraging the public sector to use and directly invest in AI applications.

The strategy also includes promoting the ethical and responsible use of these technologies, e.g., by developing standards that ensure transparency and fairness.

In August 2024, the National Science and Technology Council (Conselho Nacional de Ciência e Tecnologia, CCT) launched the Brazilian Plan for Artificial Intelligence (PBIA) to guide the development and ethical, sustainable deployment of AI systems in Brazil. With an expected investment of R$23 billion (US$4.12 billion) over four years, the PBIA aims to make the country a global leader in innovation and efficiency in the use of AI, particularly in the public sector.

What is Brazil’s approach to AI regulation?

Brazil was one of the first countries in Latin America to begin considering the regulation of AI technology. However, legislators have not yet reached a consensus on the best approach to protect rights while encouraging innovation. Bill No. 2,338/2023, referred to as the “AI Legal Framework”, adopts a comprehensive, cross-sectoral approach to AI regulation.

AI Legislative Timeline in Brazil

Similar to the European AI Act, the latest version of the AI Legal Framework adopts a risk-based approach to regulating AI, focusing on the risks associated with the use and implementation of AI systems.

If approved, it will mandate that AI developers and deployers classify the system’s risk level before its market introduction or use. Consequently, the obligations of developers, distributors, and deployers, as well as individuals’ rights, will vary depending on the AI system’s risk level.

  • AI systems that pose an “excessive risk” will be prohibited. This includes, for example, systems that employ subliminal techniques harmful to health and safety, exploit vulnerabilities of specific groups, assess crime risks, involve lethal autonomous weapons, or are used by the government to evaluate, classify, and rank individuals based on social behavior.
  • For high-risk AI systems, the AI Legal Framework categorises them based on their use cases, including those employed in critical infrastructure such as traffic, water, and electricity management, as well as essential services and sectors like justice, healthcare, education, employment, and more. This also extends to biometric identification and immigration/border control systems.

The AI Legal Framework applies to all AI systems except for those used by natural persons for private and non-economic purposes, developed for national defence purposes, and used in testing, development, and research activities for scientific investigation.

The development, implementation, and use of AI systems must adhere to principles such as human participation in the AI cycle and effective human supervision; trustworthiness and robustness of AI systems; information security; due process of law; the right to oppose AI system decisions; traceability of decisions throughout the AI system life cycle; prevention, precaution, and mitigation of systemic risks from intentional or unintentional use of AI systems; and non-maleficence and proportionality between the methods employed and the legitimate purposes of such systems.

Additionally, AI developers, distributors, and deployers must implement governance measures to ensure transparency for individuals or groups affected by such systems. For high-risk systems, developers must document the system’s functioning and design decisions, conduct reliability tests, use appropriate and representative training data, record the degree of human supervision, implement measures to mitigate discriminatory biases, and ensure explainability, among other requirements.

Individuals or groups affected by AI systems will have the right to privacy and data protection; to human involvement (e.g., for verification of output), depending on the system’s risk and the state of technological development; and to protection from illicit and/or abusive discrimination, as well as the correction of direct, indirect, illegal, or abusive discriminatory biases.

In the case of high-risk systems, individuals will have the right to explanation, review, contestation, and human oversight of decisions, recommendations, or predictions made by the AI system.

What structures are used to invest in the Brazilian AI sector?

Brazilian AI investments typically utilise standard venture and growth financing structures, often involving a Brazilian limited liability company (Ltda.).

Foreign investors can invest directly or through holding companies, often based in Delaware (U.S.) or the Cayman Islands; this is done for tax and operational efficiency.

Common instruments include equity and convertible notes, with valuations based on comparable transactions and financial projections. Compliance with foreign exchange regulations and tax considerations are crucial.

What are the key approval requirements for investing into the Brazilian AI sector?

While Brazil lacks specific FDI controls for AI and generally maintains an open policy towards foreign investment, some existing mechanisms could indirectly impact investments. Sector-specific regulations in areas like telecommunications and defence may impose ownership restrictions, and although rare, government intervention is possible in transactions that raise national security concerns.

Currently, there are no specific AI-related merger control requirements in Brazil. However, general merger control rules still apply, and transactions involving AI companies may be subject to review by CADE if they meet the relevant turnover thresholds (e.g., one party with BRL 750 million/US$ 132 million or more and the other with BRL 75 million/US$ 13 million or more of turnover in Brazil) and have a potential impact on the Brazilian market. While there are no specific regulations targeting AI mergers, CADE’s analysis may consider the potential impact of AI technologies on competition in the relevant markets.

In Brazil, although there are no specific AI-related export controls, general export control regulations still apply, particularly for dual-use or sensitive technologies.

What characterises Brazil’s regulatory landscape for AI technology?

AI regulation is still in flux

Brazil was one of the first countries in Latin America to discuss AI regulation.

The debate on AI regulation in Brazil began in early 2019 with the publication of the Brazilian AI Strategy. However, legislators have not yet reached a consensus on the best approach to protect rights while encouraging innovation. The legislative process for the Proposed AI Legal Framework is likely to be delayed due to the upcoming local elections in the second half of 2024.

Strict Accountability & Liability

Once adopted, the Proposed AI Legal Framework will establish a fault-based liability for AI agents. In consumer relations cases, however, liability will be strict.

Sanctions for non-compliance may include:

  • Fines of up to BRL 50 million (approximately US$9 million) per violation.
  • Fines amounting to up to 2% of the total net turnover of the infringing entity’s group in Brazil for the last fiscal year.
  • Temporary or permanent suspension, partial or total, of the development, supply, or operation of AI systems.

Existing obligations apply to AI system providers and deployers

Although there is no comprehensive AI law yet, existing regulations still impose significant obligations on AI system providers and deployers.

Existing Obligations for AI Systems

Paulo Marcos Rodrigues Brancher is a partner at Mattos Filho and Distinguished Professor at Pontifícia Universidade Católica de São Paulo.

Sources

  1. Reuters, 11 June 2024. Brazil hires OpenAI to cut costs of court battles
  2. BusinessWire, (2022), Semantix Announces Effectiveness of Alpha Capital’s F-4 Registration Statement, Last accessed 25 July 2024
  3. Tractian, (2023), Retrieved from: https://tractian.com/en/blog/tractians-raises-45m-in-series-b, Last accessed 25 July 2024
  4. Contxto, (2022), Retrieved from: https://contxto.com/es/brasil/healthtech-startup-alice-series-c-softbank/, Last accessed 25 July 2024