Quick Answer:
Brazil's Senate approved the Artificial Intelligence Legal Framework (PL 2338/23) on December 10, 2024, and the text moved to the Chamber of Deputies. In December 2025, the government sent Congress a complementary proposal creating the National AI System (SIA), to be led by a new council (CBIA) alongside ANPD.
Key Takeaways:
- Senate approval: Brazil's Artificial Intelligence Legal Framework (PL 2338/23) was approved by the Federal Senate on December 10, 2024, according to ClarkeModet, and moved to the Chamber of Deputies for review.
- Risk classification: the framework distinguishes between excessive risk, prohibited in certain cases, and high risk, which requires algorithmic impact assessments under Articles 13 through 28, according to ClarkeModet.
- New Brazilian AI Council: in December 2025, the government sent Congress a proposal creating the National AI System (SIA), to be led by the Brazilian Council for Artificial Intelligence (CBIA), with representatives from up to five ministries plus ANPD, according to DPL News.
- ANPD's central role: the authority would set transparency rules, impact-assessment requirements, and incident-reporting procedures, plus accredit auditors and monitor high-risk AI use inside the federal government, according to DPL News.
- LGPD-style penalties: the original framework provides for fines of up to BRL 50,000,000 per infraction, modeled on Brazil's General Data Protection Law, according to ClarkeModet.
The bill that would define how Brazil regulates artificial intelligence continues to move through Congress. The Federal Senate approved the Artificial Intelligence Legal Framework, identified as Bill No. 2338/23, on December 10, 2024. Since then, the text has remained under discussion in the Chamber of Deputies, while the government has continued sending complementary proposals to define how the new oversight structure would work once the law takes effect.
From Senate Approval to the Chamber of Deputies
According to ClarkeModet, an intellectual property firm that tracks Brazilian legislation, with the Senate's approval on December 10, 2024, the text moved to deliberation in the Chamber of Deputies and, if approved there, would be sent for presidential sanction. The bill was prepared by the Internal Temporary Commission on Artificial Intelligence (CTIA), which consolidated a text aimed at balancing protection for people affected by AI systems with the regulatory obligations of so-called AI agents.
ClarkeModet identifies three types of agents under the framework: developers, distributors, and implementers. Developers are responsible for the creation, design, and maintenance of AI systems. Distributors market or mediate the sale and supply of AI systems that have already been developed. Implementers are the entities or individuals who apply AI systems in their own operations, integrating them into their processes or services. Each of these agents carries specific and distinct responsibilities under the framework.
How PL 2338/23 Classifies Risk
The Artificial Intelligence Legal Framework adopts a risk-based approach. According to ClarkeModet, AI systems are classified into categories such as excessive risk, whose use is prohibited in certain cases, and high risk, which requires algorithmic impact assessments under Articles 13 through 28 of the bill.
Obligations by agent type, according to ClarkeModet:
- Implementers: maintain documentation on accuracy and safety, mitigate discriminatory biases, ensure human oversight, remain transparent about system operation, and conduct impact assessments at every stage of the AI lifecycle.
- Developers: record governance measures, conduct safety testing, mitigate discriminatory biases, provide information to implementers, and promote social responsibility and sustainability.
- Distributors: ensure the regulatory compliance of marketed systems, provide clear information about functionality and risks, facilitate communication between developers and implementers, and ensure systems are properly registered and certified where applicable.
According to ClarkeModet, general-purpose and generative AI systems carry additional obligations aimed at preventing risks specific to that type of technology.
The National AI System (SIA): Who Would Coordinate It
The original bill establishes the National System for Regulation and Governance of Artificial Intelligence (SIA), coordinated by the National Data Protection Authority (ANPD), supported by the Permanent Council for Regulatory Cooperation in Artificial Intelligence (CRIA) and the Committee of Experts and Scientists on Artificial Intelligence (CECIA), according to ClarkeModet.
A year later, in December 2025, the Brazilian government sent Congress an additional bill to give that structure a more defined institutional form. According to DPL News, that proposal creates the National System for the Development, Regulation and Governance of AI, which would bring together ANPD, ministries, and sector regulators under a single arrangement meant to unify AI-related policies and rules.
Per DPL News, the new SIA would be led by the Brazilian Council for Artificial Intelligence (CBIA), a collegiate body made up of representatives from up to five ministries plus ANPD itself. The CBIA would be tasked with setting strategic guidelines, proposing regulatory adjustments, and guiding the actions of the other bodies involved, and could also recommend revisions to the lists of use cases considered high risk.
ANPD would remain the central piece of the model. According to DPL News, in addition to issuing general transparency rules, impact-assessment requirements, and incident-reporting procedures, the authority would take on regulation of any areas lacking a dedicated specialized body. The proposal would also let ANPD accredit institutions for auditing and research work, publish annual reports, and monitor high-risk AI applications used by the federal administration.
Sector agencies, for their part, would retain authority to regulate the technology within their own markets. According to DPL News, each sector authority could set its own rules, require designated technical officers in certain cases, conduct audits, and sign agreements with developers to resolve disputes or regulatory uncertainty. The proposed SIA would also include two advisory committees — one for civil society and industry participation, and another made up of independent experts — intended to help shape policy and broaden technical dialogue.
Development Incentives, Penalties, and Copyright
Beyond the regulatory piece, the December 2025 proposal also opens space for development-support measures, according to DPL News. These include research initiatives, incentives for developing domestic AI models and solutions, workforce training and requalification programs, and measures to mitigate automation's impact on the labor market. Per the bill's stated rationale, as cited by DPL News, the government's goal is to prepare the state to coordinately handle growing AI adoption and to complement the debate already underway in the Chamber of Deputies over the technology's legal framework.
Penalties, according to ClarkeModet:
- The Artificial Intelligence Legal Framework provides for penalties similar to those under Brazil's General Data Protection Law (LGPD)
- Fines of up to BRL 50,000,000 per infraction
The bill also addresses copyright as it relates to AI. According to ClarkeModet, the text provides that rights holders may prohibit the use of their content, or demand remuneration, when that content is used to develop artificial intelligence systems. ClarkeModet sums up the bill's scope by noting that it represents a significant regulatory advancement that places Brazil at the forefront of AI regulation, while also posing challenges for companies and developers to adjust their practices to the new requirements.
What This Means for Your Business
While Brazil formalizes how artificial intelligence must behave within its borders, businesses across Latin America and the United States face a parallel shift: the AI engines your own customers already use to search, compare, and decide are rewriting the rules of digital visibility. A regulatory framework like PL 2338/23 confirms something already evident in the market — artificial intelligence has stopped being a novelty and become infrastructure that governments feel compelled to regulate, and that your customers already use every day to make buying decisions.
If your business is not optimized for ChatGPT, Perplexity, or Google's AI Overviews to mention it when a potential customer asks about your service category, you are invisible to a growing share of buyers, regardless of which country you operate in. MerchandisePROS's AI Search Optimization (AEO) service evaluates exactly those signals — the same ones that determine whether an AI system confidently names your business — and delivers a concrete plan to close that gap. You can review the full detail of this and other services on our services page.
"Brazil's AI regulation is one more signal that governments already treat artificial intelligence as critical infrastructure. Businesses that do not adapt to how AI presents them to their own customers will fall behind, law or no law."
- Diego Medina F, Founder of MerchandisePROS
Frequently Asked Questions
What is PL 2338/23, Brazil's AI Legal Framework?
It is the bill that establishes Brazil's Artificial Intelligence Legal Framework. The Federal Senate approved it on December 10, 2024, and the text moved to the Chamber of Deputies for review, according to ClarkeModet.
How does the AI Legal Framework classify risk?
According to ClarkeModet, the framework classifies AI systems into excessive risk, prohibited in certain cases, and high risk, which requires algorithmic impact assessments under Articles 13 through 28.
What is the National AI System (SIA)?
It is the structure that would coordinate AI regulation and governance in Brazil, bringing together the National Data Protection Authority (ANPD), ministries, and sector regulators, according to DPL News.
Who would lead the new National AI System?
Under the proposal the Brazilian government sent to Congress in December 2025, the SIA would be led by the Brazilian Council for Artificial Intelligence (CBIA), made up of representatives from up to five ministries plus ANPD, according to DPL News.
What penalties does Brazil's AI Legal Framework include?
According to ClarkeModet, the framework includes penalties similar to those under Brazil's General Data Protection Law (LGPD), including fines of up to BRL 50,000,000 per infraction.